Tube Tester design: High Voltage analog switch / mux, and other aspects

[TinkeringSteve]

Edit: Changed subject from just HV switches to tube tester..., as I'm also interested in comments besides the switches.
I might build one (and I do know an EE who would take a final look, ensuring I'm not doing something dangerously stupid).
There are quite some manual switching & measuring tester projects out there for me to know what, and how, to measure.
But I'd rather not do manual switching of pinouts etc.


----
Howdy!

I found this little bugger here:
https://docs-emea.rs-online.com/webdocs/166e/0900766b8166ed01.pdf

Do things like that exist also in forms which can withstand 400V (really, not abs max)?
Actually some 1-to-8 analog mux with fewer pins than an array of 1:1 swichtes would be even better.


Background:
I have looked at vacuum tube tester projects recently.

There are those manual ones, either with a set of heavy duty "chickenhead" rotary switches to map different pinouts of various tube types onto a measuring/testing circuit's terminals.
Or wilder constructions with lots of banana jacks to establish the connections with cable salad. (my least favorite variety.)
Then I saw a project with a bazillion relays, doing everything in an automated way.
Mmmmmh, I like automated.

Attention, now (latest) begins non-EE crazytalk, of a big kid who likes to tinker...
...but who has built tube radios as a kid, from 1950's books circuits (but in the 90's), which worked, AND survived to tell about it!

So I thought, hey, instead of several bucks per switching unit of something like 9 x 7 = 63 units (assuming noval socket, and mapping wires to a "virtual" pentode socket of 7 pins for measuring, where I assume arbitrary pinouts for valves - maybe it can be less because there are pinout permutations which do not exist... I'm "familiar" with only a couple of tubes).

I saw that there are MOSFETs which can do 600V and 3A and such, at 50cents/piece at low volume purchase - much cheaper than relays. Wouldn't that be something.

Then I thought, hey, are there maybe HV analog mux ICs, and I found the linked switch IC, but none that can do 400V.
For most tubes I know so far I need at least 250V, and I saw other projects going to 400V, so I guess it'd be a good idea to cover that.
That would be really sweet, instead battallions of single MOSFETs, eh?

----
Edit:
Other aspects of such a tester - so far I heard suggestions for:

• generation of anode and grid HV using a Royer Converter like Fig.4 here, and replace the regular pot with a digipot: https://www.analog.com/media/en/technical-documentation/application-notes/an118fb.pdf
• measuring the anode current with this kind of IC: http://www.ti.com/lit/ug/tidu833/tidu833.pdf
• an microcontroller shall be the master of the whole thing, switching things according to selected type, measuring stuff with ADCs, maybe check for user errors (e.g. selecting wrong type) by doing "all" possible tests before engaging high voltage, e.g. check whether anode and cathode pins are not swapped with the filament pins and other un-nice scenarios; record characteristic curves, compare them to database of what should be (or show closest match for tube where type is not readable on the outside) and whatever you can think of (and I feel like implementing, lol)

----

Comments?
Products?

[Kleinstein]

I don't know any cheap high voltage mux chips. The linked chip is kind of high voltage already. However it has quite some leakage (e.g. 1 µA typ) and may thus not be that good for higher impedance signals.

If it needs to be solid state, photomos is my best bet. Those bare MOSFETs would need 2 for a switch and quite some extra circuit for the control - so not a simple way.
It is still possible to find some reasonably cheap relays (e.g. $1 range)

There are likely not that many possible pin-outs. So no full cross switching needed. Some lines may also be interchangeable in the tester (e.g. different grids).

[TinkeringSteve]

Those bare MOSFETs would need 2 for a switch

You mean in a push-pull kinda way?
Is that really necessary if I place a certain type of MOSFET on a target pin of the "virtual mapped-to tube socket", according to the expected direction of current flow?
I.e., I'd not expect a current ever flowing the reverse way into the anode of a tube?
(pardon my ignorance, I've built many a circuit, but am not exactly intimately familiar with MOSFET usage)

[BrianHG]

What about something like this: ASSR-5211-001E
https://docs.broadcom.com/docs/AV02-0152EN

Or a cheaper lower power : TLP170J(F)
https://toshiba.semicon-storage.com/info/docget.jsp?did=324&prodName=TLP170J

If you want to pay, you can go 5x the current of the first guy, like: CPC1984Y a 1amp opto analog switch.
http://www.ixysic.com/home/pdfs.nsf/www/CPC1984Y.pdf/$file/CPC1984Y.pdf

Here is a mid range 600ma device: G3VM-601CR, but costs more than the 1 amp device.
https://omronfs.omron.com/en_US/ecb/products/pdf/en-g3vm_cr_fr.pdf

Note that your analog switch IC is only +/-100v at the IOs, or 0c to 200v.  And the on resistance is around 25 ohms.   The ASSR-5211-001E part I listed on resistance is 2.5 ohm in DC mode 0v to 600v, or 5 ohm for AC/DC mode -600v to +600v isolation.

400v optos are cheaper, but, the leakage current as you approach the maximum voltage begins to jump potentially to the max rating in the data sheet.  If you keep under 350v, then search the 400v parts for cheaper devices.

[David Hess]

Those bare MOSFETs would need 2 for a switch

You mean in a push-pull kinda way?
Is that really necessary if I place a certain type of MOSFET on a target pin of the "virtual mapped-to tube socket", according to the expected direction of current flow?
I.e., I'd not expect a current ever flowing the reverse way into the anode of a tube?
(pardon my ignorance, I've built many a circuit, but am not exactly intimately familiar with MOSFET usage)

A single MOSFET will work if you know the current will only flow in one direction however what about under fault conditions like a damaged tube?

I would go with using power MOSFETs driven by photovoltaic optoisolators also.  The major problem will be drain-to-source leakage through the body diode.  The datasheet specifications are worst case and are more from testing requirements than actual performance but I suspect if you qualify the MOSFETs yourself, you can find some with a leakage of 10s of nanoamps or better.

[amyk]

Tubes are high voltage low current devices (except for the filament) --- I would use relays, because then there's a far lower chance of a bad tube destroying the tester.

[BrianHG]

Tubes are high voltage low current devices (except for the filament) --- I would use relays, because then there's a far lower chance of a bad tube destroying the tester.

Miniature latched telecom relays would do the trick (optional latching type allows a single power shot, then no power needed from your digital logic supply making super low power draw overall).  Today, from China, these are dirt cheap as well.  Though, a proper design with the opto-mosfets I mentioned above with current limited supply or series resistors to prevent maximum current of the mosfet switch devices.

[T3sl4co1l]

PhotoMOS relays may be of interest.  Guessing the capacitance and leakage won't be a problem (compared to a mechanical relay).

Doing it with discrete MOSFETs, you'll need two in series, source to source, with an isolated gate driver.

You may want to handle heater connections separately, maybe with mechanical relays.  The combination of on resistance and voltage standoff is not so semiconductor friendly.

Tim

[Ian.M]

Switching 400V isn't so easy, even if you can do so at zero load current, as you need a reliable >>400V isolation rating between contacts.

It may be worth looking at rotary switches driven by RC servos.  You could do the switches 'multimeter style' on your PCB, with gold plated contact pads direct on the PCB and a common arc or circle, the servo mounted direct to the PCB on the other side, with ts shhaft through a hole and a multi-finger phosphor bronze moving contact on a 3D printed arm on the servo shaft. Silver plate the moving contacts or solder on pure silver sheet on the tips before slitting them.  That would make it easy to do N way switching with enough clearance and creepage distance for 400V.   Shut down the HT and heater supplies before moving the servos to minimise contact erosion and inter-contact tracking across the insulation.

[TimNJ]

I don't know anything about this project, but Steven on the Macrofab podcast just finished building a tube tracer.

http://analogeng.com/2018/09/11/utracer-3-upgrade-part-2-design-overview/