|
Components and Circuits For discussions about component types, alternatives and availability, circuit configurations and modifications etc. Discussions here should be of a general nature and not about specific sets. |
|
Thread Tools |
28th Aug 2020, 10:57 am | #1 |
Hexode
Join Date: Mar 2018
Location: Bishop Auckland, County Durham, UK.
Posts: 373
|
Measuring gm of a valve without a valve tester
I have quite a few ECC81 and ECC83 valves and want to test them for emission and gm by making simple DC measurements. I was thinking of wiring a valve holder into a circuit to supply say 100volts to the anode and a variable negative supply to the grid with the cathode to HT negative. I could set the grid voltage to, for example, 2v negative then measure the anode current. Could I work out the gm by changing the grid voltage by one volt and then measuring the change in anode current? If the HT was fed via a resistor would this complicate things as a change in current would mean a change in anode volts? I just want a rough check, would this arrangement work?
__________________
Regards Martin |
28th Aug 2020, 11:41 am | #2 |
Dekatron
Join Date: Apr 2011
Location: Cornwall, UK.
Posts: 13,454
|
Re: Measuring gm of a valve without a valve tester
For a mutual conductance test the anode voltage needs to remain reasonably constant.
Lawrence. |
28th Aug 2020, 12:57 pm | #3 |
Octode
Join Date: Nov 2010
Location: Ross-on-Wye, Herefordshire, UK.
Posts: 1,648
|
Re: Measuring gm of a valve without a valve tester
As ms660 notes, the anode voltage should stay sensibly constant - so don't feed the anode via resistor. Use a stabilised supply from a bench supply if possible. Otherwise you have the principle bang on.
I can see why you want to alter the grid voltage by 1 volt - to keep the maths very simple - but in practice you just need a change in anode current and grid voltage that you can reasonably accurately measure. In other words, if you only have an analog current meter (e.g. AVO) which is 10mA FSD, then try and go for a change in anode current which is sensibly large on that scale - e.g. 1mA or more. If you try and read a change of say 100uA on a 10mA FSD meter, you will get serious inaccuracy because you don't have the resolution. Richard |
28th Aug 2020, 2:04 pm | #4 |
Dekatron
Join Date: Apr 2009
Location: Oxfordshire, UK.
Posts: 4,924
|
Re: Measuring gm of a valve without a valve tester
I've prepared ECC81's for use as "standard" valves by doing what you are describing. The one problem you have watch out for is the valve starting to oscillate, which will make your readings worthless and an ECC81 will do that very willingly! One of the main things to do prevent oscillation is to put a few ferrite beads on the leads as close as possible to the valve holder.
I think I used a higher anode voltage than 100V; I used the test conditions specified in the Avo Databook, so that the results which I got enabled direct comparison with valves tested on an Avo. Let me know if you need any more info on that. B
__________________
Saturn V had 6 million pounds of fuel. It would take thirty thousand strong men to lift it an inch. |
28th Aug 2020, 2:28 pm | #5 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,803
|
Re: Measuring gm of a valve without a valve tester
Which gm do you want to measure?
It changes continuously across variations in bias conditions. The AVO valve data books just give the nominal value at one specific spot set of bias conditions. It is fine for its purpose as a simple single-point test. You then have to assume that if it is OK, then the rest of the family of curves is OK. When the valve is not being tested, but doing its day-job, it is being moved all over the family of curves by the signal, and different circuit designs may start it from different initial conditions. So you really need the family of curves to be within expected bounds. A single point test is likely as good as you're going to get. Mostly it works. But beware of AVO data for some valves. Big valves, ones where the VCM ca't reach the normal operating voltages and currents. For these, AVO quote Ia and Gm at levels substantially below their normal working area. You have to assume that if things are within AVOs pass/fail band at these reduced conditions, then the rest of the curves must be OK. This leaves you open to valves that exhibit limited breakdown voltages, or limited saturation of cathode emission. They'll test good, but not work for the applications which need them. At the small signal radio set end of the spectrum, the pass fail tolerances of the testers are a lot narrower than many sets will work fine with, so failed valves according to testers seemed just fine. So, take valve test results with a pinch of salt, unless someone like David Simpson or Bazz has plotted a set of curves on a DC tester. It all serves to make the prices paid for VCMs a bit of a joke. I want a cartoon drawing.... someone is at an auction bidding silly money against someone else for a late-model VCM while a tall guy in robes points at them shouting "Thou shalst not worship graven images!" while ushering in a busload of fundamentalist nutters. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
28th Aug 2020, 3:33 pm | #6 |
Dekatron
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,061
|
Re: Measuring gm of a valve without a valve tester
I'd do it with AC.
100mV p-p on the grid, from a small transformer with a 1kΩ potentiometer across it (you could use the heater supply. A 1kΩ load resistor, which will hardly change the anode voltage. Scope the grid, tweak for 100mV p-p (or a convenient figure). Scope the anode, see the voltage p-p here, which is numerically equal to the current p-p in mA. Then, one easy calculation! |
28th Aug 2020, 3:55 pm | #7 |
Dekatron
Join Date: Sep 2007
Location: Oxfordshire, UK.
Posts: 4,310
|
Re: Measuring gm of a valve without a valve tester
It's not a problem for small-signal valves (at least I don't think it's a big problem, compared with the ones David's already mentioned) but for power valves you have to be a bit careful with quasi-DC measurements of gm that overall temperature changes, caused by changes in anode dissipation, don't lead to cathode emission changes which can contribute to changes in Ia. If you use kalee20's AC approach you'll keep the average temperature constant and that problem will go away.
In a valve like the ECC83, with a very low gm, you might also need to make sure that there isn't any other AC noise getting into the detection circuitry which could otherwise confuse your measurement. Cheers, GJ
__________________
http://www.ampregen.com |
29th Aug 2020, 1:27 am | #8 |
Hexode
Join Date: Mar 2018
Location: Bishop Auckland, County Durham, UK.
Posts: 373
|
Re: Measuring gm of a valve without a valve tester
Thank you for your comments and suggestions. I don't have a 'scope so would have to make do with DC checks. Rather than use a meter set to current range, I was going to wire a 1K resistor in each anode circuit and measure the voltage dropped with a digital meter. When used in audio amp circuits, the ECC83 seems to run with 60 to 100v on each anode and Ia of 0.5 to 1ma.
__________________
Regards Martin Last edited by martin.m; 29th Aug 2020 at 1:29 am. Reason: typo |
29th Aug 2020, 9:49 am | #9 |
Dekatron
Join Date: Sep 2007
Location: Oxfordshire, UK.
Posts: 4,310
|
Re: Measuring gm of a valve without a valve tester
Vak for the ECC83 can be higher than that - it's 137V in the Mullard 5-10 and 180V in the Leak Stereo 20 phase-splitter (somewhat less in the input stage). In the second stage of the Quad 22 it is indeed 100V but in the third stage it's 200V.
EDIT: I have contemplated an AC design using a small output transformer connected 'in reverse' in the test valve's anode circuit i.e. with the low-impedance side between anode and HT. The step-up action should give a larger AC voltage for easier measurement without putting a significant impedance into the anode circuit and should also provide handy DC isolation between the measuring device (I was thinking of a DMM on an AC range) and the high-voltage HT. The downside, of course, is that the set-up would need to be calibrated. Cheers, GJ
__________________
http://www.ampregen.com Last edited by GrimJosef; 29th Aug 2020 at 9:56 am. |
29th Aug 2020, 10:09 am | #10 |
Rest in Peace
Join Date: Mar 2010
Location: Berkshire, UK.
Posts: 72
|
Re: Measuring gm of a valve without a valve tester
It is also worth noting at there may be quite a wide variation of gm at a specific operating condition. For example the Brimar 6067 (special quality version of the 12AU7, ECC82) gives a design figure of 2.2m/V with acceptance limits of 1.75 to 2.65mA/V for a specific anode and grid voltage.
Anybody who has used the wartime VR65 (SP61) and VR91 (EF50) will be aware of the range in perfectly good valves. Chris |
29th Aug 2020, 10:38 am | #11 | |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,803
|
Re: Measuring gm of a valve without a valve tester
Quote:
Not really do-able with ordinary multimeters, though. Good application for that Brookdeal stuff on the surplus market that no-one knows what to do with! David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
|
29th Aug 2020, 10:41 am | #12 |
Nonode
Join Date: Mar 2008
Location: Aberdeen, UK.
Posts: 2,839
|
Re: Measuring gm of a valve without a valve tester
Thread responders have mentioned AC i/p's to the Grid & measurement of corresponding AC on the Anode. That isn't (static DC conditions)Gm, its Dynamic Gain "G". Basic static formula for a valve is mu = ra x Gm. For the rest of valve formulae - look up "Barkhausen's Law". The maths involved for dynamic conditions require the inclusion of Anode Load Resistance - Ra. The likes of Iliffe Valve Data Books & others have page by page tabulations for Gm, ra, & Ra, to aid design folk, as well as basic recommended valve electrode voltages.
AVO's range of VCM manuals show a basic recommended circuit for DC Standardisation, using external(stabilised, or constantly adjusted) DC PSU's. Using the ECC range(or their equivalents) of valves can be risky for the un-initiated. The wee things(some jolly expensive) have a tendency to go "flash, bang, wollop" if Vg is inadvertently reduced too much, or Va increased too much. They "don't like it up em", excessive Ia wise. Regards, David Regards, David |
29th Aug 2020, 12:07 pm | #13 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,803
|
Re: Measuring gm of a valve without a valve tester
As the AC on the grid is reduced, the AC on the anode will also reduce and conditions will tend from dynamic towards static.
Using a lower value anode resistor will also help get closer due to smaller anode swing. For an AC measurement, this can be compensated with some gain. Can we get close enough to static conditions? yes. Can we check that we have? also yes. Measure Gm, then do it again with even less AC on g1 and see if you get the same result. Keep on reducing the AC amplitudes until it makes no difference, and then you aren't there (you never will be) but at least you are so close that it makes no difference. This is a reality check that works in numerical integrations, calculus and many other fields. If using a spectrum analyser or network analyser, try slowing down the sweep and seeing if the picture changes. If it's the same, then you are sweeping slowly enough. So speed it up until you see a change, then bring it back one step. If you have a family of curves and want to extract Gm at some bias condition, you have to draw a tangent to the appropriate curve at the appropriate point. Now, ask yourself how you drew that tangent.... you looked away from the centre point and tried to balance the errors didn't you? But you needed to not look too far out because the different curvature on either side spoils the balancing act. Same thing as the AC technique! So it all comes out in the wash. Swings and roundabouts. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
29th Aug 2020, 3:53 pm | #14 |
Nonode
Join Date: Mar 2008
Location: Aberdeen, UK.
Posts: 2,839
|
Re: Measuring gm of a valve without a valve tester
As David says, the calculations involved at any one wee point (dIa/dVg) on the overall DIa/DVg curve can get a bit complicated maths-wise, seeing as the curve is a function of x sq'd, more or less.
Basically, the Gm plotted curve gives circuit designers a working point of suitable grid bias at which the signal passing through the valve encounters the least distortion & benefits from the best amplification. Audio amp. designers drool over valves with matching Gm curves. As do speculative internet valve wheeler/dealers. Regards, David |
29th Aug 2020, 3:59 pm | #15 |
Dekatron
Join Date: Apr 2011
Location: Cornwall, UK.
Posts: 13,454
|
Re: Measuring gm of a valve without a valve tester
If you do AC you need DC bias anyways so might as well go DC and measure the change in anode current, doing AC with zero bias might get you up the wrong slope.
Lawrence. |
29th Aug 2020, 6:47 pm | #16 |
Dekatron
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,061
|
Re: Measuring gm of a valve without a valve tester
Agree with Lawrence, you do need bias! However, doing with DC you might turn to zero and shoot up the Ia.
Doing with DC, you also have drift to cope with - and you're only dealing with a few hundred μA change anyway unless you change Vg a lot, and if you do that, you're changing the operating point. I'd stay with the low-amplitude, low frequency, AC. As long as the load resistance is much lower than the valve's published ra, the error due to changing Va will be only a few percent. |
29th Aug 2020, 6:57 pm | #17 |
Dekatron
Join Date: Apr 2011
Location: Cornwall, UK.
Posts: 13,454
|
Re: Measuring gm of a valve without a valve tester
My test would be to crank up (set Vg & Va) for near max anode dissipation and see if Ia agrees with the valve data, if not make a note of Ia and make a judgement call.
Lawrence. |
29th Aug 2020, 8:38 pm | #18 |
Dekatron
Join Date: Apr 2009
Location: Oxfordshire, UK.
Posts: 4,924
|
Re: Measuring gm of a valve without a valve tester
From the OP, I think this is just a sorting exercise to identify good/poor/useless valves and if that is so, I think that KISS applies.
My own experience with ECC81 DC test results was that when subsequently tested on my VCM163, the results were in very pleasing agreement. I tested them under Avo databook conditions, but swung the Vg a little bit either side of the specified value to produce a line against Ia, five data points up then down. When David Simpson organised the 'Round Robin' exercise, generally, the various 163's which participated in that all showed good correlation with David's test results (I cannot remember which valve type he used). I drew DC curves for a few different valves to use as Standards for the 163. I started with ECL80's (which someone had been recommended on a website) because they are plentiful but fairly useless and you get two valves in one envelope. I finally settled on ECC81's because if you sort through a few you can usually find a valve which has an Ia test result just below 10mA and a Gm just below 6, those values being Full Scale readings on the 163's first range, and so still show usable readings on Ranges 2 and 3. B
__________________
Saturn V had 6 million pounds of fuel. It would take thirty thousand strong men to lift it an inch. |
29th Aug 2020, 8:53 pm | #19 |
Nonode
Join Date: Dec 2012
Location: Kirk Michael, Isle of Man
Posts: 2,346
|
Re: Measuring gm of a valve without a valve tester
A few years ago, i made a simple box with a valve holder, flying leads and a heater transformer which I connected to my CT71 curve tracer. i forget exactly how i did it, but just looked at two curves on one vertical gave instant GM. This was for ECC8x valves, dead easy and it seemed to work. All now sold on. I was going to make a simple plug in adaptor for testing my box of nuvistors, but never got around to it.
Les. |
30th Aug 2020, 11:21 am | #20 |
Nonode
Join Date: Mar 2008
Location: Aberdeen, UK.
Posts: 2,839
|
Re: Measuring gm of a valve without a valve tester
Bazz, they were 6AQ5's & 6AU5's along with their Ia/Vg(Gm) A4 Graphs. Many Forum folk joined in the "Round Robin" exercise, which involved about 20 ish valves which are still out there in the fraternity. I didn't want them back & hoped that folk have made them available to others.
Right enough, VCM163's gave close results(so they should at £1k plus), as did a few CT160's & their MK3/4 siblings, as did a couple of Sussex's. Martin in Sweden joined in, & his RoeTest results were a confirmation boon. Like Martin, a number of participants used 21st century software to produce graphed curves. Old doddering analogue me just produced hand drawn curves. My "join the dots" simple graphing produced a series of wee straight lines which made up the exponential curve. But under magnification, one could see much smaller "straight lines" on some of the computer generated curves. These wee "straight lines" are what David & I would call "deltaVg", i.e. a very small linear change in grid voltage, and in fact when one twiddles a mA/V knob on a valve tester - that's what we're measuring. Maths & Physics experts will tell you that there are an infinite number of dx's in all such graphed curves. Delta Vg - a large change in grid voltage, defines in many cases the extremes of an applied ac voltage & resultant undistorted Ia & subsequent ac o/p Va. So, to test heaps of ECC valves - build a wee rig as per the AVO Manual's design, supply it with steady DC for all the valve's electrodes, and Bob's your uncle. Don't fanny about with AC, and for God's sake start out with a higher Vg than the book value, then gently reduce it whilst monitoring the rise in Ia . Both Va & Vh must remain spot-on their book values. 2nd hand working LV DC PSU's often pop up in Forum or BVWS or VMARS sales/auctions/etc. - for supplying Vh & Vg. HV DC PSU's can be made from re-purposed(I love that term) redundant radio or test equipment circuitry. Regards, David |