A Dashing Enigma

Did you know this site is now 7 years old? Time sure flies. GDEMU itself is older yet by about 2 years, the first version that I’ve made was based on Altera Cyclone II dev kit and mostly worked but had some timing issues. Initially I had no serious intentions of selling it – though a year later I figured maybe a small self-contained PCB would sell two dozen or so units, and would help me raise some money for other projects. Little did I know how it’s going to snowball. Well, at least I can say the project goal was achieved – even if I don’t like how ODEs became the new modchips and everybody expects to get one for a couple bucks to play “free games” (and I’m the bad guy for not mass-producing ODEs to flood the market, forcing low prices).

Anyway, obviously the last post was an April Fools’ effort. The photo was of a dirty 5.25″ double-density TEAC floppy drive, I deal with those only in personal projects and that particular one is now clean and working. Testing it revealed issues with two of my 386 mobos – the one on which I’ve replaced a broken crystal resonator with a tad bit faster device, and is now overclocked, which is nice but also screws up system timers. I need to undo this, the extra 1.x MHz is not worth these issues. Another mobo, and this one was badly corroded from NiCd battery spill, is now acting up – randomly hangs on DMA transfers (so mostly in games that use SoundBlaster voice playback and while using floppy drives). I suspect there is a corroded via under the 82C206 chip but there is no guarantee the mobo will survive desoldering it – there’s already pad damage and wires around to fix affected traces. This mobo was, until now, my “daily driver” for testing 386/486 code but I think I will replace it with a P200 MMX and maybe attempt the fix later. It served its purpose and I have a few other mobos for quick code testing if I need a true 386 system for that.

And speaking of testing 386/486 code, I have collected some performance data from various FM Towns machines using my own benchmark program. Frankly there’s tons of that data but in order to show some of it in a simple, easy to digest way I’ve made one chart that I think best reflects the actual overall CPU performance.

FM Towns - Performance

First, let me introduce the machines present on the chart:

  • A – gen1 (MODEL 1), 16MHz 386DX
  • B – gen2 (2F?), ~18MHz 486DLC
  • B1 – B with cache enabled
  • C – gen3 (40H), 16MHz 486SXL
  • C1 – C with cache enabled
  • C2 – C with cache enabled, clock x2
  • C3 – C without cache, clock x2
  • D – gen4 (CX), 16MHz 386DX
  • E – gen4 (CX), 16MHz 486SXL
  • E1 – E with cache enabled
  • E2 – E with cache enabled, clock x2
  • F – gen2 (2F?), 16MHz 486DLC (*)
  • F1 – F with cache enabled (*)
  • G – gen2 (2H), 16MHz 486SXL
  • G1 – G with cache enabled
  • G2 – G with cache enabled, clock x2
  • G3 – G without cache, clock x2
  • H – Marty, 16MHz 486SXLC
  • H1 – H with cache enabled
  • X – Fresh-ET, 66MHz 486DX2

(*) The F machine used a slightly different version of the benchmark executable and 386/486 CPUs are really sensitive to code position, so over many loops the result can be a few % off (faster or slower, depending how the code changed).

I’ve also added some colors to try and make this more readable. Black is 100% standard machines, blue is 386 CPU replaced with 486 DLC class. Orange is same as blue but with CPU clock resonator replaced as well (there are some serious limitations to this technique and even small overclock can cause issues in these systems). Violet is SXL/SXLC CPU with clock doubler enabled.

So, what can we see here? First of all, the black bars – the gen1 and gen4 machines have identical performance in default settings. The Fresh-ET is a bit faster in the compatibility mode but not by much – it mostly equals a CPU swap to a 486DLC (which all of the towers can do, except generations 3 and 4 require desoldering the old one first). I guess it’s not really possible to perfectly match a 386 performance with a 486 chip on a twice faster system bus. That being said, based on my tests with DLC chips, this is not enough to make a difference if the game doesn’t support FAST mode.

The “fast” in the chart is either FAST mode (for machines that support it) or reduced RAM wait states (as much as it was possible). This offers a decent speedup, in general reducing WS on 386 tower machines gives comparable results to replacing the CPU with 486 DLC – though obviously if you can replace the CPU and then also reduce WS, it’s even better. But if you want to keep your machine 100% original then WS reduction (with ODE) is probably the way to go if you need some extra performance. Even gen1 tower, which can only reduce the WS by 1, shows some speedup.

A 486 DLC with WS reduction or FAST mode can cut the benchmark time almost in half, and keep in mind this offers pretty much perfect game compatibility. You can get even better results by enabling the CPU cache (also via ODE) but that will break some games – some will work fine, some might have audio or video glitches, or even unresponsive controls. It looks nice on the chart but in reality can be quite a lot of trouble. This also goes for any overclocking, as mentioned above. And then there is clock doubling on SXL chips, this has some effect even with no cache and just WS reduction – which is nice, considering the issues with cache, but without having cache enabled you gain maybe a few extra % and that’s it. So it’s not really worth paying extra for a 486 SXL over DLC, unless the price is similar.

I didn’t test x2 clock on Marty because it breaks a lot of things for some reason (tower models don’t suffer from these issues). For example it can completly screw up mouse emulation on gamepad, something that Marty does by default via its BIOS. Not sure about other 386SX based systems, maybe UG series would benefit more thanks to having a native 20MHz clock in FAST mode. Then there is the proper x2 clock doubler in Fresh-ET, with built-in cache coherency protocols, burst mode and faster system bus. A true DX2 class CPU is more than twice faster than SXL and has none of the issues – so if you really need a fast Towns machine for some reason, get a late Fresh (or at least a true 486 desktop).

To sum this up, with Towns the 486 is faster than 386 but only in FAST mode (and most games don’t use it). Even older machines can be “tuned up”, either by swapping the CPU or forcing FAST/WS reduction via ODE. Or both. This is often enough to make Street Fighter 2 playable without major slowdowns for example.

Good Optical Drive

Are you tired of small, noiseless and reliable drive emulators? Looking to impress friends with the oldest, dirtiest computer on the block? If so I have something just for you!

Good Optical Drive 1

Announcing a new device – the Good Optical Drive, or GOD in short. You can now restore your machine to it’s old electromechanical glory with the slowest, most power hungry motors I could find.

  • Nostalgia factor x100 from the smell of old dust and tobacco smoke alone.
  • You get to feel old just by looking at it and knowing what it is.
  • Live in Texas? This will double as house heater come next winter storm.
  • Watch the street lights dim and go out as you power it on.
  • Finally a proper load for that 6MVA military surplus diesel generator you bought.
  • The sweet, sweet whine of dry, rusted bearings – no moving parts were ever cleaned or lubricated.
  • Old, yellow and brittle plastics that will snap off right after unpacking.
  • Makes a quick and simple tester for your house smoke detectors (*).
  • And best of all, it sometimes even works and does not destroy the disc!

(*) actual fire not guaranteed.

GOD will be available in limited quantity on first come, first served basis. Get your GOD now!

Project Time! #5

WordPress decided to finally block the classic editor for good – which makes me very unhappy. This block editor is a disaster, maybe useful if you are one of those my-phone-is-my-life hippies but I guess I’m too old for that. I use my phone for making calls, and not even of the video variety. So as I’m writing this 70%+ of my PC screen is a white, blank space with a small column of text somewhere in the middle, and I can barely see the font. If it wasn’t for FireFox 140% page size I’d be struggling to write anything at all. It’s “progress” like this that annoys me the most so don’t even get me started on Windows “evolution”.

It’s not my intention to turn this into old people rant, I just wanted to warn you that from now on things might look different, and not in a good way. And I dread the next ordering window, I used to be able to just copy-paste the forms directly into page code, I have no idea how to do that now.

Anyway, originally I wanted to make a whole post about choosing the right kind of J-FET transistors for the input switches in meters like Fluke 8600A – especially since the ones they’ve used are not made anymore (and that’s been true already in late ’90) and difficult to find. Usually your only source is another meter as a part donor, or maybe ebay if you don’t mind that a few of those transistors would cost more than what the whole meter (in working conditon!) is worth. And you’re very likely to get an expensive fake anyway. But a very technical post like that would be boring for most people, and with the latest WP changes I don’t even feel like doing it anymore. So I’m just going to mostly wrap this up with a few more photos.

The inputs – a very important part of the meter, you are going to plug and unplug the jacks pretty often, it’s a good idea to have a fuse for current measurement, and – in general – you might be applying high voltages to these so there better be proper clearances and insulation there. Well, if you are a Fluke fan then I’d suggest you never open one of these 8600A, ignorance is a bliss they say. I’ve opened a cheap phone charger once and promptly dumped it into a nearby recycle bin shortly after, even though it was still working – I would not dare to connect it to AC outlet in my house anymore. Anyway, I had to rewire the Fluke inputs because:

  • Clearances, what are those? Surely a thin layer of old plastic and/or half a milimeter of air gap between bare metal parts will handle 1200V AC just fine.
  • Let’s just use a thin, strained wire to connect ground returns to PCB, that will obviously not just snap off on it’s own after years of use.
  • Insulated fuse holder? Like, an actual fuse holder, not a cheap short metal tube on one end? What are you, a safety loving socialist?
  • This here single spring blade will hold the fuse in place just fine 95% of the time. And the fuse will not move to sides and make a poor contact when you turn the “holder” to lock it in, not at all.
  • If it works, it must be a Fluke!

This is how it looks now. Trust me it’s a major improvement over what I found:

FLUKE 8600A inputs

The range-selecting GAL with the capacitor update I mentioned before, and a peek at my first attempt to rebuild the power supply. The AC converter board is still out but everything DC almost worked by this time (BTW note the input fuse, the fabled Fluke quality and safety, it’s why the meter was so expensive):


Unfortunately my idea of using efficient DC-DC converter to create 5V for the logic didn’t quite work out. Even though 5V is not used in the A/D converter, there was enough noise present on the PCB to upset it. Ignore the extra wires going to the rectifier bridge, I was using that with external DC power supply because I didn’t want to have 230V running throught the board (right up to the ON/OFF switch in the front panel) while I was working on this.

FLUKE 8600A power

I could not get a zero reading in lowest DC ranges with this setup, so sadly it had to go. I ordered a custom 7,5V/1A transformer for this project and went with linear 5V regulator. But I was stubborn and decided to keep the +/- 18V DC-DC converters since those feed 15V regulators so there shouldn’t be much noise. And actually the battery-powered option from Fluke uses a similar setup, but there’s a bit more EM shielding. I can’t replicate that here though, no space for it – the battery option has an extra PCB (and a different transformer BTW) which would prevent me from installing the data output unit that I want to have.

Since I had a curious issue with the ohms converter not being able to calibrate – that is until I swapped one of the factory-selected precision resistors for a different value – I decided to create my own PCBs to replace the burned ones. I wanted to know if the carbonized PCBs are the cause (some leakage currents perhaps) and also I was not entirely sure connecting more than 100V input would be safe anymore, considering all the damage to the input divider. Plus it turned out one of the relays on the divider was on it’s way out anyway (yup, that unwrapped one). My PCBs are not 100% identical to the originals but work just as well, if not better – while reusing only the custom parts and semiconductors, the rest is new.


FLUKE 8600A Ohms

This is how it looks installed, with the updated PSU and new transformer:

FLUKE 8600A new PCBs

This transformer is a split-bobbin type but I decided to add a bit more mains RF rejection by running the input wires through a ferrite bead, as well as adding two 100nF SMD capacitors next to the bridge (on the other side of the PCB) to supress harmonics. Original transformer had electrostatic shield but this should be just as good. This one is 230V only but that isn’t an issue for me.

FLUKE 8600A trafo

I was able to calibrate the DC ranges, there aren’t any noise problems except the AC converter, but that one is another story – kinda works but has issues, so I will attempt another reproduction PCB at some point I guess.

And finally the data output unit, now upgraded with USB port so that it can be used with a PC directly, or remotely via LAN+RPi. The DOU output is fully isolated and thus so is the USB. I made it in such a way that USB module is only powered over the host cable, in order to prevent any backfeed between it and DOU.


A small modification to the case was required to make sure the USB port will come through and stick out just far enough for the plug to fully lock. It was either that or a massive edge connector plug with PCB and a lot of extra chips to process all the 5V signals present on the original printer port. I think this approach is better and more importantly, it works.


Sadly, the meter only reports what range it’s in, not what mode is selected. But still it’s pretty useful, I’ve implemented a few SCPI-like commands and this is what the USB communication looks like:

Fluke,8600A,2106127,1.2 (2021-03-13)
+04,953e+0 V DC
+04,954e+0 A DC
04,953e+0 V AC
04,953e+3 Ohms
+04,954e+0 V DC
+04,953e+0 V DC
+04,953e+0 V DC
+04,953e+0 V DC
+04,953e+0 V DC
+04,953e+0 V DC
+04,953e+0 V DC
+04,953e+0 V DC
+04,953e+0 V DC
+04,953e+0 V DC
+04,952e+0 V DC
+04,952e+0 V DC
+04,953e+0 V DC
+04,953e+0 V DC
+04,953e+0 V DC

To sum this up, I’ve fixed or replaced:

  • 50 ohm pot (blown; R8 on DIV board)
  • F2591 J-FET with selected BF245C (blown; Q14 on main board)
  • F2591 J-FET with J111 (suspicious; Q15 on main board)
  • CD4053 chip (failed; U16 on main board)
  • ROM chip with 16V8 GAL (failed; U9 on main board)
  • 2,7M and 4,7M resistors (drifted way above 5%; R74, R82 on main board; R17, R18 on DOU)
  • wire connections on input jacks (broken; on main board)
  • factory-selected resistor with 1% metal-film of higher value (something drifted; R90 on OHMS board)
  • new power supply section of my design (missing parts; on main board)
  • fuses for input and power supply (missing parts; on main board)
  • new PCBs (charred; DIV and OHMS boards)
  • reed relays along new PCB (suspicious; DIV board)
  • 470nF film capacitors (open/short; C14, C20 on AC board)
  • 150k resistor (drifted above 5%; R43 on AC board)
  • 15k resistors (drifted above 5%; R32, R33 on AC board)
  • 25nF ceramic capacitors (cracked; C6, C7 on AC board)
  • FD700 diode (missing; CR16 on AC board)
  • case screw (missing)

The case screw was actually a serious problem. It’s an imperial thread (and it comes in coarse/fine variants too!), kind of difficult to find in Europe.

The Wretched Automatons, contd.

Just letting you know DocBrowns will be available this Saturday (2021-02-20), and this will be another USA-friendly window opening at 20:00 CET. This particular batch is a small one so I might limit it to Marty owners only – we’ll see how it goes.

There will be a short break from mid-March to early April as the Post Office here will introduce some changes to stamps and payment for international shipping. At the same time we’ll also be testing a different internal workflow – this is related to changes in VAT procedures that will soon apply in EU. In general nothing that you should worry about except there now might be a delay (a few days) between your payment and actual shipping taking place. That’s because, for tax related reasons, we won’t be able to ship first and then do the paperwork later as we did so far.

UPDATE: DocBrown orders are now closed. Give me a day or two to process the orders and send out confirmation emails. Just eyeballing the queue and it looks like I should have enough devices for all orders – I hope I’m not short just a few.

UPDATE 2: I have a small batch of Phoebes that are ready to go so I will open orders this Saturday (2021-03-06) at usual time of 12:00 CET. I hope to have most of these shipped before the April break.

Project Time! #4

Last time I showed a Hewlett-Packard 3465A multimeter that I bought pretty cheap. It was described as powering up but misbehaving and I got it for one reason – to do science to it.

For some time now I wanted to build my own “multimeter”, though that’s a bit of a stretch as far as the name goes – it would probably only do DC voltages but maybe also resistance if I could come up with a good enough current source. Autoranging because actually it’s easier these days to have an MCU or some logic drive a couple of relays than to design/find a complicated gang switch. And if I could get my hands on suitable precision shunts then DC current measurements would also be possible. I did not exactly expect to build a quality instrument, no, rather it was supposed to be a learning exercise.

Now, one could just get an ICL7106 and build a 3.5 digit meter out of that and a few extra parts (like the 7-segment displays) but that’s not how I roll. I wanted to make my own dual slope A/D converter, with (MOS)FET switches, voltage reference and all the control logic including auto-zero, averaging and digital calibration – the whole works. Eventually though I figured that maybe trying to fix a broken meter rather than building one would be a better idea considering my goals and lack of experience.

The HP multimeter was promising but ultimately a failure – in the sense that all I did was take it apart, remove the dead NiCd cells, clean it, flood the switches with contact cleaner, put it back together and it worked. So well in fact that I was impressed and decided not to tinker with it. It’s a really good device – I actually ended up using it quite a lot.

So, back to square one, I got me another cheap multimeter – a Fluke 8600A. I’m not very fond of Flukes even if some people consider them to be the superior brand. Obviously this is 40+ year old tech and times (and quality) have changed but comparing the guts and performance of both HP and Fluke I have to say the HP wins, hands down. Here’s the seller’s photo:

and a link to EEVblog forums post that will show you how the Fluke should look inside.

Well, mine has a few issues here and there:

Power filtering electrolytic capacitors are gone. So is the 5V rail regulator along with its heatsink. Also, and that’s the major pain in the butt, so is the transformer – all missing. Worse yet, whoever removed the voltage regulator was not very proficient with the soldering iron:

Turns out the very pin that connects the regulator to GND is also the common ground tie point between digital and analog sections of the meter. So with the via all but destroyed I had pretty high impedance between DGND and AGND and it took me quite a while before I found on the schematic these two should be in fact connected.

Well, I can supply the power (+5V, +15V and -15V) from external PSU for now. But what’s this, the metal shield around AC to DC converter is missing as well. Also, I see some burn marks:

Not exactly sure how it happened but the precision wirewound resistor in the ohms converter burned up, and somebody replaced it with what seems to be a metal film 1% resistor. But I have to say they did a pretty good job because originally it was a 9.95K 5W 0.1% – good luck finding something like that. The replacement is 9.96K which is good enough except maybe temperature coefficient but not much can be done about that. Same for the PCBs but at least the carbonized material around the leads has been removed. Here’s what the add-on cards look like up close:

The voltage divider on the photos is already missing the 50 ohms calibration pot, I removed it because it was open. Possibly went kaput in the same event as the ohms converter? Didn’t have a spare on hand so for now I just put in 2 resistors just to be able to continue testing. You probably noticed the missing markings on one of the reed relay coils, somebody was messing with that one too.

Meter had problems with autoranging so the first thing I suspected was the small ROM that’s used to drive relays – it’s known to die. That turned out to be mostly good, one bit was glitched in the 200mV AC range. Some people claim this part is actually a bipolar PROM, not mask-ROM, which would make the zero-bit recoverable with a suitable programmer (unlike modern EPROMs, bipolars have all zeros by default and are programmed to ones by burning connections inside the chip). That was however moot point as I don’t have such programmer, and shortly after one of the outputs died – could be I killed it by shorting with probe tip at some point.

That’s the ROM testing setup, since it has only 32 addresses I can just test all of the values in 10 minutes by switching the inputs with wires. The output that died is that single LED on top, the photo and notes I made prove it was working up to a point.

The fuses were missing (both the line power and the current measurement protection), and the whole thing would pretty much always show overload no matter what mode or range was selected. This was eventually traced to busted FETs switching input and zero to the A/D. In general I had to:

  • rebuild the power supply
  • find and replace faulty logic chips
  • fix issues with A/D converter
  • clean input jacks area and redo wiring
  • fix broken AC to DC converter
  • replace burned PCBs (and dying reed relays in the process)
  • recalibrate everything (which required replacing a selected resistor)

I’m going to get into the details of all this eventually, for now though I’d like to present my idea of replacing the faulty ROM. I’ve seen people use a more modern EPROMs like 2764 (which is also obsolete now but at least obtainable and mostly supported), but that is 28-pin wide body part. The only way you can fit that in is on wires, that’s ugly.

With some minimal pin swapping I’ve managed to use GAL16V8 instead because the contents of the ROM is simple enough that it can be turned into logic equations that fit on the part. And the GAL is small enough that it actually will fit into the original socket:

The results were promising but two issues came up. First one was some weird latching in invalid state when switching away from DC current mode that caused the GAL to basically short-circuit the 5V line. Good thing I had this running from external PSU with current limits. This particular problem was easily solved by adding a 10uF ceramic SMD capacitor right on the chip, as close to power supply pins as possible. I have to say Fluke apparently did not belive in any logic chip power filtering, there’s basically nothing of the sort on the whole PCB and the 2-layer layout means pretty long traces. So long in fact that the usual 100nF wasn’t cutting it, 1uF almost fixed it completly, so I picked 10uF to be on the safe side. That and the whole issue with MLCC caps derating under DC bias.

Second problem was also DC current mode related – the way Fluke implemented it was by cutting power to the ROM and all the relays while it was selected. This is because current measurement is still manually ranged on these early models, by selecting the correct shunt resistor, and the A/D is permamently set to 200mV range to keep the burden voltage low. Now this power cutoff worked fine with ROM becuase it has open-collector outputs that in turn drive PNP transistors (that in turn drive the relay coils). For that to function properly the transistors have pull-ups to 5V that shut them off when ROM is not driving the base low. GAL however is a modern part with push-pull CMOS outputs protected by diodes to VCC and GND. So those resistors would backfeed 5V to the GAL, via those diodes, providing it with enough power to actually function and drive some of the transistors when it shouldn’t. So I had to add a mod, a small NPN transistor that would isolate the GAL from GND when the power was cut. This has a minor downside of shifting the GAL ground up by some 0.2V in the on state, but that’s not an issue in this circuit. See if you can spot the tiny SMD part (note no external base or bias resistors, these are built into the transistor package):

This solution works very well and is almost a drop-in replacement, except for that small mod. It’s not a big deal though as it will also work properly with the original ROM in the socket – which was one of my goals (in case I’ve managed to revive the ROM somehow).

BTW, the reason I got that Fluke and not something else in the first place was this add-on card:

This is opto-isolated digital output board that sits over the main PCB, I have plans to try and build an interface between it and a Raspberry Pi, which would allow me to turn this experiment into a networked logger. A dumb one since this board only provides output and can’t affect the meter operating mode or range, but that is still pretty useful to me.

The Wretched Automatons

Phoebes are done so I will open orders for GDEMUs this Saturday (2021-02-13) at 12:00 CET. Then, in a week or two – depending on how fast I can ship those GDEMUs – I will open orders for DocBrowns. I have a small batch of those almost ready.

News are slow lately but I have some non-ODE related side projects that might interest some of you, though I’m trying to figure out if it’s even worth mentioning the ones that I’ve little to show for. But there is one that I consider “done”, even though I might just revisit it soon. But it served it’s main purpose – which was to teach me a lesson 🙂 More details after the weekend.

UPDATE: GDEMU orders are now open closed. Give me a day or two to process the orders and send out confirmation emails.

Somnolent, contd.

You will be able to order Rheas during this Saturday (2021-01-23). Let’s make it USA-friendly time since I haven’t done it in quite a while, so the orders will open at 20:00 CET.

Orders from UK are allowed – for now at least. This new VAT system is not cool though. It’s one thing if EU does it on a common platform for all its countries and it’s free and has the same rules in every country. If we have to pay monthly fees for legal access to UK market, and have to keep up with all the changes in the UK law and procedures from now on. then it’s not really worth it for small operation like ours is…

UPDATE: Rhea orders are now closed. Confirmation emails will be sent out in a day or two.

UPDATE 2: I will open Phoebe orders this Saturday (2020-01-30) at usual time (12:00 CET).

UPDATE 3: Phoebe orders are now closed. Confirmation emails will be sent out in a day or two.


I will take orders for GDEMU on Saturday 2021-01-02 at noon (12:00 CET) as usual. I expect this to be a short run and after that it’ll be either Rhea or Phoebe next.

I’m also considering advance orders for Wizards. Long story short – I will probably treat this ODE differently and prefer modders to buy a few pieces at once rather than do individual sales. This is because it takes some soldering skills to properly install Wizard and it’d be even better to have a proper crimping tool to do all the wires nicely, with connectors, rather than solder them directly to the PCBs. Plus the size of the Wizard makes it difficult to ship it as cheaply as I have done so far. Basically there’s lots of small issues completly unrelated to the ODE itself that are making this one more difficult. Packing for shipping alone is so time consuming…

Due to the whole brexit thing I will not accept orders from UK at this time. This is temporary restriction – I wish to avoid the initial hiccups and have my packages delayed, bounced or taxed at the border. Once all the new customs procedures are well in place I will be happy to resume shipping. So, probably in a month or two?

At the same time I remind everyone else that I will cancel orders that I can’t ship due to any unforseen COVID-19 restrictions on travel (which tend to make air shipping impossible). Hopefuly though there won’t be any more of these and year 2021 will be less crazy.

UPDATE: GDEMU orders are now open closed. Orders from UK are conditionally accepted and might be delayed until further notice. Please allow a day or two to build the list and send out confirmation emails.

Project Time! #3

What can one do with a piece of vintage tech like HP 6920B? I found a few uses for it already:

  • testing “magic eye” tubes that require some 250V for anode/screen voltage
  • testing capacitors for DC leakage up to 1kV
  • reforming electrolytic capacitors
  • testing gas-discharge voltage-regulator tubes
  • testing Geiger–Müller radiation detectors
  • testing electroluminescent backlight panels
  • testing transformers (ratios, self-heating and winding insulation)
  • and zapping stuff with 1kV AC/DC

I’m probably going to invent some other uses for it as well but originally I wanted it for some repair fun and one particular purpose – as a 1kV DC source to recalibrate my AimTTi 1908P multimeter:

First let me introduce the 1908P a bit. I got some other AimTTi gear as well so I expected their multimeter to be decent. Different people have different needs so here’s my personal pros and cons lists:


  • 5 and 1/2 digit modern digital multimeter with reasonable specs
  • it’s cheap, even taking into account Chinese offerings like Rigol and Siglent
  • has Ethernet port for TCP/IP connectivity (as well as serial and USB, and GPIB/IEEE-488 if you need it)
  • no fan, no noise, no holes in the case for the dust to get in
  • 2s boot-up, of which 1s is the display test
  • decent screen with good backlight
  • manufactured in Europe
  • built-in battery pack for off-line operation


  • it’s only 120k counts
  • not very fast, 4 readings per second at full resolution (less in dual-measurement modes)
  • no rolling averaging to reduce input noise
  • diode test goes only up to 1.2V due to non-selectable range even if the output voltage is about 3.5V
  • only 10M impedance on input on DCV ranges (less in AC mode obviously)
  • built-in battery pack for off-line operation

I’m really starting to dislike USB these days. I use it a lot and I’ve seen all kinds of issues with it – cable length problems, flaky hubs and connections in general, driver problems. In fact custom, broken, never updated drivers are the worst and can often interfere with other devices that worked perfectly before. So I always look for Ethernet ports on new equipment, I don’t trust hardware manufactures to come up with decent drivers or any at all if you aren’t on Windows or it’s not the version they support. Serial port is second best option but it’s slow and typically offers no HW error correction, so a single bit glitch in the data stream can easily bump a digit up or down in the reading and you’d never know it. GPIB/IEEE-488 is… old. And costly. And very proprietary. If you already use it a lot you won’t mind but newcomers are better off looking for something else.

The 1908P has Ethernet port and it’s a huge plus for me. And the price is still acceptable, unlike say Rigol offerings that give you cheaper USB-only models or the full-fat ones for 500 USD more. The 120k counts is more of “5 digits” than “5 and a half” in my book but I’m not going to argue with established shady marketing, it’s not just AimTTi doing this. I very much like the no-fan design and the LCD display is OK, not great but also not annoying me in any way. The battery pack is a bit of a double-edged sword – nice when you need it but keep in mind it will die at some point. Mine in fact did (was probably defective) and that most likely caused the brick I had when updating FW – AimTTi did good though and repaired this for me for free.

No averaging, 10 meg input and limited diode mode are not great but also not a dealbreaker. I might at some point need more than 4 reading/s at full 5-digit resolution but I don’t yet so that’s also not a problem. All in all I am happy with the price and performance – or rather, I was, until I discovered a glitch.

Long story short: When in 1000V DC range and below 100V (so only with manual ranging) there is a weird non-linear jump at exactly 99.00V that skips all the .01 to .05 values and goes to 99.06V (so from 99.00 to 99.06). I asked AimTTi about it and they suggested recalibration of the 1000V range. Turns out the calibration procedure requires a 100V input at one point, which is pretty much the only thing I’ve changed in the calibration data and it only made the “99V problem” worse. So clearly the jump/skip is some sort of FW issue that tries to improve the linearity of this range by having both 100V and 1000V references stored – but messes up the cross-over point. At least that’s how I see it. AimTTi has since ignored 3 of my emails – both about this issue and another one with their MX180TP PSU.

I’ve read many good things about AimTTi support but I have to say my personal experience with them is spotty at best. They will fix bricked equipment and issue FW updates for the serious bugs (not to brag but the Interface firmware version 1.07 for MX180TP exists because of me). The less obvious stuff though? Won’t even reply to the email.

And so, to conclude this – would I buy this multimeter again, knowing what I know now? Yes, it’s not like Rigol/Siglent or the Keysight/Keithley (which are often made by Rigol/Siglent in the lower price range anyway) have spotless record when it comes to FW bugs. And the price is still very good. But can I recommend AimTTI to others? No, I can’t. If you really need a good bench meter then you should probably look into the (considerably) more expensive K-brands. If on the other hand it’s for personal use then Chinese meters have more features, color LCD graphical screens and you probably won’t mind having just the USB available anyway. Unless you live in UK and can drive to AimTTi in person if need be.

Dewy Fields

Phoebe orders will be open this Saturday at noon CEST. New Zealand, Argentina and Vietnam are back on the shipping list.

UPDATE: And done. As usual give me a few days to process all the orders and then I’ll will send out confirmation emails.

UPDATE 2: I should be done with Phoebes this week. It’s taking longer than expected because PayPal is glitching and showing all paid requests as “Sent” even though I haven’t actually entered the tracking number yet. Long story short it’s a lot more work for me to sync PayPal with my own spreadsheets. In other news, I will be opening GDEMU orders this Saturday at noon (do keep the winter time change in mind!).

GDEMU orders are now open closed. Give me a day or two to process the orders and send out confirmation emails.