what would be the most demanding tube preamp topology?

[capacitorless]

I need to sell some vintage 12au7 and 6sn7 tubes, to pay for parts for a DIY compressor project. I have a couple of testers, with excellent gas/shorts testing and Gm (not perfect, but decent) testing covered.

What I want to do is to test for noise, hum and microphonics. So, rather than design to reduce things like hum, I'm looking for the worst case topology

Would a classic, old school, simple 2-stage cascade be the best way to go? Or would SRPP like in DIY ECC802S (12AU7 / ECC82) Vacuum Tube SRPP Preamplifier or cascode (or another topology I'm unaware of) be more sensitive to a marginal tube?

My gut says to go with a simple two-triode cascade, using good HV and non-elevated AC heater supplies & with typical plate & cathode resistors, and run that into an interface. I would then use something like REW to measure things.

Any help & advice much appreciated!

 

[ruffrecords]

When I spent a lot of time sorting out low noise and low microphony double triodes I configured them as mu followers simply because then you know the gain is going to be very close to mu but it does not matter too much what topology you use.

The main problems you will have will be eliminating extraneous noise and interference from the test set up. You definitely need to build the circuit in a screened box with the tube fitted in a screened socket. To measure tube noise and microphony you should short the input to 0V. Make sure you use a large enough output capacitor to pass any 1/f noise. If you decide to run ac heaters you biggest problem will be hum and its elimination is as much a matter of layout as anything. I would recommend using dc heaters. I used batteries on my first test rig.

Cheers

Ian

 

[capacitorless]

Thank you for the excellent advice. The top search result was The Valve Wizard -Mu Follower and it calculates the ECC82 component values, even

Did you use the upper output, or the lower (high impedance) output instead?

Some of these tubes could be considered "premium" examples, so I want to have solid specs, without sending them out for 3rd party testing. They are all vintage organ pulls, from discarded tube based organs.

I've read that such tubes can be specially screened, but I'm not going to assume so I just want to minimize returns and make sure people are happy, without giving them away, haha!

 

[ruffrecords]

I used the upper output because of its relatively low impedance. My very first tube mixer used mu followers throughout. It is still in use in a Swiss studio today.

Cheers

Ian

 

[Matador]

I built a rig exactly as Ian describes in a small metal box, inputs are heater voltage (DC) and a plate supply. It's wired as a common cathode gain stage with a 220K plate resistor, biased to roughly 1 mA with a bypassed cathode resistor. Grid is shorted to 0V, however it has a switch to send it to ground via a 1M resistor. Output is from the plate through a 0.1uF cap to a 100K resistor to ground, and is sent to a headphone amp.

You can stick in a tube, let it warm up, then rap the side with a pencil to listen to the microphonics and background hiss.

 

[emrr]

Noise may take a full day burn-in with tubes that've been resting for a long time (decades?). Larger batches of new tubes seem to need it as well, RDH4 specifies a burn-in time for new manufacturing testing.

 

[gyraf]

 

[capacitorless]

Thanks, folks, this is all super helpful

I have multiples of the same tube, so I'll start by taking a pair of (eg) RCA clear top/side getter 12au7s that test the same for gm, listen to them, and burn one of the two in for 24h. I'll listen again to the burned-in one and we'll see what's different and I'll react accordingly.

I have an old but working Moviola 6V6 SE chassis on hand with a spare noval socket. There's a DC bench supply here that will work well for the filaments, and I have enough parts around for the rest. Soldering this up while watching SB LVII.

Edit: @emrr decades for certain. Almost all of these were in old-school late 50's/early 60's organs, that were in various states of non-functionality and used tubes for the tone generation circuits.

 

[gyraf]

..note that many-many NOS stock tubes you get these days has already been burnt in, many of them several times, by manufacturers buying in-bulk, burning in and selecting the best, then selling the rest back into the market. Ever wondered why nearly all GE 5-star 6072's you come across are that microphonic?

/Jakob E.

 

[ruffrecords]

I built a rig exactly as Ian describes in a small metal box, inputs are heater voltage (DC) and a plate supply. It's wired as a common cathode gain stage with a 220K plate resistor, biased to roughly 1 mA with a bypassed cathode resistor. Grid is shorted to 0V, however it has a switch to send it to ground via a 1M resistor. Output is from the plate through a 0.1uF cap to a 100K resistor to ground, and is sent to a headphone amp.

You can stick in a tube, let it warm up, then rap the side with a pencil to listen to the microphonics and background hiss.

I love the way you can listen as the tube heats up, the hiss increases to a maximum and then falls back as the space charge develops. I also love listening when you switch the heaters off but leave the HT on. You can hear all sorts of pings and tings as the innards cool down.

Cheers

ian

 

[scott2000]

I love the way you can listen as the tube heats up, the hiss increases to a maximum and then falls back as the space charge develops. I also love listening when you switch the heaters off but leave the HT on. You can hear all sorts of pings and tings as the innards cool down.

Cheers

ian

When can we expect the album? Sounds interesting actually.

 

[Mr. Moscode]

What do you mean by most demanding?

 

[capacitorless]

I tried to explain this in the first post but probably not completely adequately. In a nutshell, I was looking to test a tube in a circuit that brings out the flaws, instead of designing for the utmost SQ.

I have some good answers and (to start) I've implemented a mu follower with a DC heater supply, as Ian suggested. None of this is super difficult to wire up. I may set it up with a switch to go with either DC or AC filaments, and isolate the power transformer, depending upon the results from initial tests.

The other thing I came up with found was the 6SN7 version of NY Dave's mic preamp, except single ended so minus the 2nd tube and output transformer.

 

[Mr. Moscode]

Thanks. I designed, built and sold about 300 Small Signal Tube Testers in the 90s. It read gain in DB, tested 6 and 12v tubes, had dc filament supply, regulated B+ and distortion test in a typical triode circuit. Having an ability to switch to ac heater would have been a good for H-K leakage.
You're on the right track.

 

[capacitorless]

I would love to be able to quote actual Gm without springing for (eg) an Amplitrex. I guess I could always resell it once I'm done with it, but this is just is a hobby & that would tie up some cash I could better use for the three bus compressor projects I have lined up, haha!

The Hickok 600 I have will give me a decent & relative Gm, but it hits a small signal tube's grid at 5vac, and I've learned that isn't the best way to go. I'm already modding the 600 with a 3 way switch to replace the English control, and a pluggable (stereo 1/4" jack) resistance reference for the Bias control, both for repeatability. So putting in an injection point & measurement output is totally cool. I'm wondering if I could inject a signal into the grid, and measure Gm using software?

Not looking to do totally scientifically accurate curves, just something that is useful to someone considering buying them.

 

[Mr. Moscode]

I rarely test tubes unless I have to match them. I put them in the circuit and check the parameters in situ since that's where it has to live. If you're going to sort and sell the bottles, I can see it. But if it's going into something you're going to use and you have bunches of them, plug them in and pick the best ones that way. You can bang on them anywhere sort it out by ear. Some that may be unusable in the first stage may be fine in a later stage depending on the gain following its place in the circuit.
Just a suggestion to help if you're in a hurry.

 

[capacitorless]

Thanks and that's a great suggestion. I'm not in a huge rush, but I do want to get started asap.

What I might do is give it my best shot at measuring Gm (with what I have) and send a batch of worthy tubes to a testing service. I would do that in one or two batches and see if I can get a quantity discount. The only ones so far that might be worth it are these late 1950s long plate Mullard/Mitcham plant Bugle Boys. They even have the treble clef logo, and painted tips which probably indicate some kind of screening was done.

I have three 12au7 that I sent out for testing, and they came back with (theoretically) accurate Gm, as well as current draw. Those numbers are higher than what the Hickok reads, by a little over 20%. For example, their tester on a strong example: 3,000 Gm (@ 8.95ma), this Hickok 600: 2,350 Gm.

If I'm understanding things correctly, the Hickok when calibrated correctly will measure an average example of a 12au7 at 2,200 Gm. I need clarification on this, but the latest I've read seems to indicate that the roll chart value is "average new" as opposed to "used threshold".

I read something yesterday that explains why some/most Hickoks test low. They said it was around 20%, and it was because it included a non-linear region around cutoff. I'll find that and post the link.

Sorry for the long post. I do testing (in the IT field) for a living, and this stuff is probably more interesting to me than most other folks, haha!

Edit: from Antique Radio Forums • View topic - Measuring Tube's Transconductance W/O a Tester

"...Because a Hickok uses AC on the plate, and sweeps over the whole operating range of the tube. If the Gm were constant over the whole characteristic curve, this method would work just as well as DC. But at the lower part of the characteristic, near cutoff, the Gm goes to zero, and this part gets averaged into the Hickok's reading.

It's not a big deal, 20% error at most, but it's there."

 

[Mr. Moscode]

The testers are often sensitive to line voltage variations. I have a TV7 which I haven't used in years but it has rheostat to adjust for line voltage before the test , , ,If I remember correctly. . . It has no DC supplies. It simply tests using AC.

 

[capacitorless]

It does and I check it before and after pressing P4 (the Gm test) and if it varied from dead center I retest

From what I found out earlier tonight, even with the AC nature of the test, and a high 5v grid signal, it's still a good test - just not able to give an accurate Gm reading. It looks like the roll chart value for Gm is "average new". The early Hickoks had values for "used threshold"; probably 70% of the "average new" value.

So based on this, I'm getting good results as it stands. The 600 results are 22% lower than the results from the outside test (I don't know what tester they used, and they are no longer in business). It all seems to jive.

I picked up a variac and isolating transformer that I had in storage, so I can check calibration now (maybe not, I might need a calibration 6L6 tube with a specific Gm under the given test conditions).

 

[Mr. Moscode]

Are you selling tubes now or is this for your Private stash?