For anyone wanting to set the PI Trim pot

Overdrive Special, Steel String Singer, Dumbleland, Odyssey, Winterland, etc. -
Members Only

Moderators: pompeiisneaks, Colossal

User avatar
alvarezh
Posts: 461
Joined: Thu Oct 18, 2007 11:11 am
Location: Santo Domingo, D.R.

Re: For anyone wanting to set the PI Trim pot

Post by alvarezh »

Nice info here guys, many thanks on my part.

All the best.
Horacio

Play in tune and B#!
thorens
Posts: 8
Joined: Wed Mar 23, 2011 3:57 pm
Location: Biebelsheim, Germany

Re: For anyone wanting to set the PI Trim pot

Post by thorens »

I really enjoyed your postings. I just wanna emphasize the importance of matched tubes (PI). If not both triodes have the exact amplification curves, the whole adjustment makes no sense. And each called adjust method is getting you closer to the perfect balance. But there is no perfect balance. At last you are trimming the anode resistors. But you have different power supplies in mornings and evenings, therefore different voltages at your phase inverter... In a youtube video Larry Carlton's tech talked about power supply problems. Amp sounds best at 122 Volts. 4 Volts less or more were critical. He was not particularly talking about the phase inverter, but the PI is affected too.
Tom
User avatar
glasman
Posts: 1446
Joined: Wed Jan 19, 2005 10:37 pm
Location: Afton, MN (St Croix River Valley)
Contact:

Re: For anyone wanting to set the PI Trim pot

Post by glasman »

Well have not visited this subject in quite a while.

Here is the procedure that I use. Does require a spectrum analyzer and a low distortion signal signal generator.

1. Inject a 100hz 0.5V signal into the PI. Do NOT use a buffered return (ie D-lator) as this makes the adjustment virually impossible. This is due to the return amp adding a lot of crap to the input of the PI. A

2. Attach one channel from the analyzer to the Negative feedback line from the 4 ohm tap.

5. On the analyzer display the spectrum and record the level of the 100hz spike, Now find the 200hz spike in the waveform and slowly adjust the trimmer to minimize this peak.

On a 100W amp the meaured level at the Feedback point is usually about 15dbV, the second harmonic an usually be tweaked to about -45 to -70 dbV.

If you find that you "run out of pot", try adjusting the one of the plate resistor values 10K or so.

You can also use a distorion analyzer to get the same results. An old HP can be had on fleabay for a $100.00 or so.

Also a balanced PI tube is the only way to go.

Also for best results, make sure that the BIAS voltage is EXACTLY the same to both output tubes. I usually check this on the 220K PI reistors.

Gary
Located in the St Croix River Valley- Afton, MN
About 5 miles south of I-94
aka K0GWA, K0 Glas Werks Amplification

www.glaswerks.com
User avatar
martin manning
Posts: 13080
Joined: Sun Jul 06, 2008 12:43 am
Location: 39°06' N 84°30' W

Re: For anyone wanting to set the PI Trim pot

Post by martin manning »

Good stuff, Gary, thanks. This procedure would get the power amp balanced precisely (at least at the test conditions) and free from the influence of any distortion coming out of the preamp. Having done this, do you find that there is nothing further to be gained by tweaking the trimmer by ear?
User avatar
glasman
Posts: 1446
Joined: Wed Jan 19, 2005 10:37 pm
Location: Afton, MN (St Croix River Valley)
Contact:

Re: For anyone wanting to set the PI Trim pot

Post by glasman »

martin manning wrote:Good stuff, Gary, thanks. This procedure would get the power amp balanced precisely (at least at the test conditions) and free from the influence of any distortion coming out of the preamp. Having done this, do you find that there is nothing further to be gained by tweaking the trimmer by ear?
My Ear Tweaks and Using test equipment are so close on the pot position, I simply use the test equipment and then test the "known" feedback points at a reasonable gain setting. Also listen for note bloom in both clean and overdrive.

Gary
Located in the St Croix River Valley- Afton, MN
About 5 miles south of I-94
aka K0GWA, K0 Glas Werks Amplification

www.glaswerks.com
User avatar
Ron
Posts: 85
Joined: Mon Aug 28, 2006 1:57 am

Re: For anyone wanting to set the PI Trim pot

Post by Ron »

I just installed a 22k pot in the phase inverter of my 50W Dumble-ish project (using a Bassman 50 chassis), with 100k resistors going to the two PI plates from the ends of the pot. I checked the 12AX7 PI tube for gain balance and the triodes are very close, and I checked the EL34 output tubes as well and they are well matched in transconductance and close in idle plate current.

I adjusted the PI balance pot using two simple but different techniques that require a scope and distortion analyzer. These different techniques gave the same result. A good digital multimeter could be used instead of the scope, so the PI pot adjustment can be made accurately without exotic test gear.

First method (same as the one by bluesfendermanblues above): I adjusted the pot so that the peak-to-peak voltages from each of the PI triode plates was identical. Voltages were measured at the EL34 input grids to avoid high DC voltages. I noted where the pot was adjusted to compare later with the second technique.

Second method: Using an HP 8905 Audio Analyzer, I adjusted the pot for minimum audio distortion at 1kHz with the amp running at a few different output levels into a dummy load (Weber Mass).

As far as I could discern, both methods resulted in the same setting of the PI balance pot. If you have a gain-balanced PI tube and well matched output tubes, then using a digital AC multimeter to measure the peak or rms outputs of the PI tube triodes (instead of a scope, since most people don't own one) is a valid way to adjust the PI balance.

Following up with ear testing and final adjustment is the last step.
Deblacksin
Posts: 1
Joined: Sun Oct 22, 2017 3:05 pm

Re: For anyone wanting to set the PI Trim pot

Post by Deblacksin »

How should such a ttimpot be implemented in a jcm800 based amp?
I modified iys preamp to ODS specs but things arent as silky smooth as tbey should be..
User avatar
sonicmojo
Posts: 669
Joined: Thu Mar 03, 2011 3:41 am
Location: Oahu, HI

Re: For anyone wanting to set the PI Trim pot

Post by sonicmojo »

Deblacksin wrote: Fri Nov 17, 2017 12:07 am How should such a ttimpot be implemented in a jcm800 based amp?
I modified iys preamp to ODS specs but things arent as silky smooth as tbey should be..
Look at some of the layouts in the Dumble files section, that should give you the idea. I don't think it will help you with the "silky smooth" though. I think you may need to look elsewhere.
---------
Bryan
User avatar
Reeltarded
Posts: 9926
Joined: Sat Feb 14, 2009 4:38 am
Location: GA USA

Re: For anyone wanting to set the PI Trim pot

Post by Reeltarded »

Change to similar values (120k/110k) use 5 or 10k trmmer, double your tail resistor (Marshall has lil tail), set voltage to around +7 volts on positive side and then finish it by ear.

I am adding it to my Bandmaster with slightly different values.

Part of the Marshall "thing" is a hot PI and the small tail for big NFB availability.

In modifying your Marshall did you go plate drive and kill the CF off? If so, this is the next step. The mixing matters, the bright matters, the tone stack matters. Dumble is a wide and dark animal compared to a Marshall with just swapped cathode RCs.
Signatures have a 255 character limit that I could abuse, but I am not Cecil B. DeMille.
Sone Aura
Posts: 21
Joined: Mon Mar 30, 2015 12:43 pm
Location: USA

Re: For anyone wanting to set the PI Trim pot

Post by Sone Aura »

An imbalance can be good a thing too, as the proper imbalance can add some nice distortion and you really need to use both your ears AND AN O SCOPE, and you "NEED" to know exactly what kind of distortion that O Scope is showing you. You also need that amp up to operating temperature before you start twiddling around with any trim pots. With tube radio transmitters, stable tuning was not attainable for up to an hour or sometimes even more. A distortion meter? Yeahhhhhh-NO!! A THD meter is nice to have, except for one glaring issue. It knows how to give a percentage reading or maybe show a dB ratio, but it doesn't know odd order from even order, and crossover also shows up. So a THD meter is good for fixing stereos, and transmitters, but it's kind of a 3rd wheel luxury item for guitar amp work. (I own one made by HP, and I use it, but just not very often on guitar amps anymore, as in almost never.) That is where the scope and your ears really come into play. Crossover distortion is mud, even order is added sustain and a subtle rounded compressed sound with octaves & 5ths, and odd order is going to add 3rds, and adds an aggressive texture and timbre distortion. A lot of that is subject to personal tastes, but I don't think anyone wants any Dumble to sound like murky mud, so we do not want crossover or intermodulation distortion. You also want to warm the amp up well so everything has settled into where the amp is going to be operating thermally after a good 20 or 30 minutes of playing, and I mean put a sheet of metal, or at least a sheet of cardboard with aluminum foil glued to it over the chassis to keep the heat in, until you are ready to go into it and dial in your tone on any trim pots, including an absolute final bias setting. If set that pot with the amp cold, it's going to drift as the amp heats up. A good book I can recommend is, "The Oscilloscope Handbook Volumes 1 and 2" by Rufus P. Turner, and any other O Scope books from the golden age of tubes and Germanium transistors. Rufus' book has a section on audio amps and wave forms, that are broken down into simplistic charts of basic waveform shapes that are super easy to interperate. It is also without a lot of crazy useless old analog television wave forms that no one cares about anymore. You do NOT need a 1GHz $50,000 scope. Any old Kikusui, Hitachi, or GW Instek 2 channel 10-50MHz scope is fine, and try to find one with a BNC jack on the back labeled "Z" axis, or "Z" axis input. That can show you phasing issues, and other cool useful stuff, and the SSS amps are going to be chock full of some cool waveforms with phase angle leads and lags, especially with that LCR tone filter set up in there.
Only man can transmit total darkness at the speed of light.
Sone Aura
Posts: 21
Joined: Mon Mar 30, 2015 12:43 pm
Location: USA

Re: For anyone wanting to set the PI Trim pot

Post by Sone Aura »

I also very highly recommend buying a decent set of "alignment tools", which were a staple item in any TV repair technician's kit. They are either nylon, fiberglass, or some other durable non-conductive plastic material. DO NOT BUY AN OLD USED SET. Digital TVs have been around for so long now that these old sets are aged brittle plastic that will just break, and make you mad that you wasted your money on them. You also don't want the risk of touching that metal, and a live trace, shorting between traces, or touching the metal and inducing noise into the amp, all of which can and does happen with screwdrivers.
Only man can transmit total darkness at the speed of light.
User avatar
martin manning
Posts: 13080
Joined: Sun Jul 06, 2008 12:43 am
Location: 39°06' N 84°30' W

Re: For anyone wanting to set the PI Trim pot

Post by martin manning »

ayan wrote: Mon Aug 06, 2012 10:37 pm After having tried the "Ampeg" method on many amps with great results, it has become my favorite method for setting the PI trimmer. You should have 1-ohm cathode resistors at the power tubes, otherwise this method will not readily work. You will need a signal generator, a dummy load, and a digital multimeter. This is my adaptation of it.

1. Tie together the cathodes of the power tubes that are in phase with each other (i.e., the tubes that have their plates connected to the same B+ wire coming off the OT's primary).

2. Connect a digital multimeter set to DC Volts (yes, DC!) to each pair of cathodes, and set the scale to millivolts. You will be measuring the voltage difference between the two pairs of cathodes.

3. Inject a 1 Vrms 100 Hz signal into the amp. I keep the amp in clean mode with the preamp volume turned down low and the tone controls where I normally set them (it will not matter much where you set them anyway, all I try to do is to not overdrive anything).

4. With a load connected to the speaker jack, set the output level (master volume) to the point where the amp dissipates about as much power as you think it dissipates when you normally use the amp. My rule of thumb is that at gigs I may be running at 50W tops, so into an 8-ohm load I will have a 20 Vrms output (P = V x V/R, so V = SQRT (PxR) = SQRT (50 x 8.) = 20). If you play your amp at lower volumes you may want to shoot for say 20W, which translates into 12.7V, etc. The output level will change the results/readings.

5. The multimeter's reading should be on the order of a few millivolts. Turn the PI trimmer until the reading is 0 mV; 0 may go by quickly, so look for the reading to shift from + to - or vice versa. When 0 is reached, each pair of cathodes is at the same voltage with respect to ground and the two sides are therefore balanced. If the reading doesn't cross over, try another tube in the PI slot; it could also mean that your power tubes are not matched.

This method does not take into account any further imbalances introduced by the OT but has, per my experience, consistently yielded excellent results. I hope it works for you too.

Gil
I'm going to reprise this method as the best way to go about setting the PI trimmer to get to an accurately balanced power stage. Some may prefer a slight unbalance, but balanced is an excellent place to begin. Perhaps most important is that this technique measures the dynamic balance at the output tubes, as opposed to the PI output or the approximate static PI plate voltage difference method. It therefore takes into account any mismatch in power tube average Gm side-to-side. You don't need a scope, just a signal source, a DMM, and a dummy load.

A few comments:

For the signal source, a phone app will work nicely, feeding signal into the amp's input using a Bluetooth audio adapter. The Ampeg (SVT) instructions mentioned above specify a 40Hz signal, but I don't think that's too critical, and maybe something in the mid-range (400-1000Hz) would be more appropriate for guitar.

Begin by setting the static bias level and balance (by swapping tubes from side-to-side to get the closest total current match).

Regarding step 2, measuring DC voltage across the power stage cathodes (the voltage difference from one side to the other) combines the sensed cathode currents in the same way as they are combined in the output transformer, so the meter is seeing a complete sine wave. It displays average voltage, which is adjusted via the trimmer to a zero value so that the full sinusoid is centered around zero volts. The available range of adjustment with a 10k trimmer will likely be less than 10 mV.

Regarding steps 3 and 4 above, I think it's important to set the balance with the output power high enough to drive the power tubes into cut-off, so that the power stage is using only the top halves of the PI output signals. Keeping the output signal undistorted and at about half of rated power seems like a good target (and that is what the Ampeg procedure results in for an SVT). Using the clean channel at a moderate preamp volume setting, and turning up the master volume to achieve half power is one method; another is to drive the power amp directly through the power amp input, which requires around 2 Vrms. [edit: IMO driving the power amp directly is the best option, since the preamp output is likely to have some second order harmonic distortion] Output power is found by measuring the voltage across the load (Vrms), squaring it, and dividing that result by the load resistance, so target Vrms = (P x Rl)^0.5, where P is output power and Rl is load resistance.]

[edit: Using a USB powered Bluetooth receiver connected to my phone, and running a signal generator app producing a 400Hz sine, I measured the receiver output at 1.5V RMS with my DMM, which should be sufficient for the purpose here. I also looked at the output on a scope, which confirmed the voltage and frequency measurements, and which showed a good quality sine wave. With that level of input a the PI, I'd expect about 18W into a resistive load for a 50W ODS power amp, 36W for a 100W amp.]
Last edited by martin manning on Sat Mar 02, 2024 5:52 pm, edited 6 times in total.
sds1
Posts: 20
Joined: Sun Aug 06, 2017 10:53 pm

Re: For anyone wanting to set the PI Trim pot

Post by sds1 »

This thread should be unpinned, that's great that everyone has come up with their own creative ways to trim the PI, their own opinions on the purpose of this calibration, but Dumble had a very specific outcome in mind for this trimpot and a very specific calibration procedure to go with it, and it's buried in this thread among other answers which are not challenged, the bad info is spreading more than the good and thus nobody knows about the Ampeg procedure, or understands how besides the point it is to calibrate the phase inverter to itself when there's a feedback loop, tubes, etc to consider.

Anyhow... Please. Unpin.

The Frenchie videos are wrong too.

Respectfully.
User avatar
martin manning
Posts: 13080
Joined: Sun Jul 06, 2008 12:43 am
Location: 39°06' N 84°30' W

Re: For anyone wanting to set the PI Trim pot

Post by martin manning »

sds1 wrote: Tue Nov 09, 2021 2:00 pmDumble had a very specific outcome in mind for this trimpot and a very specific calibration procedure to go with it, and it's buried in this thread among other answers
I don't think unpinning it is the right answer. Can you point to the specific procedure that goes to the designer's intent? Is that the Ampeg procedure?
sds1
Posts: 20
Joined: Sun Aug 06, 2017 10:53 pm

Re: For anyone wanting to set the PI Trim pot

Post by sds1 »

martin manning wrote: Tue Nov 09, 2021 2:21 pm
sds1 wrote: Tue Nov 09, 2021 2:00 pmDumble had a very specific outcome in mind for this trimpot and a very specific calibration procedure to go with it, and it's buried in this thread among other answers
I don't think unpinning it is the right answer. Can you point to the specific procedure that goes to the designer's intent? Is that the Ampeg procedure?
Yes, the Ampeg SVT procedure.

A thread with that procedure at the top, with consensus that it is consistent with Dumble's design and intent, followed by a bunch of "here's how I do it" would be much more beneficial to the community.

A recent discussion on Facebook reveals the general builder public seems to be aware of every PI trim method BUT the Ampeg SVT procedure. And the discussion veers into a very generic "PI plate balance" (100k/82k) discussion which is just not very relevant here.

This site is well-known as the primary reference for all things Dumble, a pinned thread implies a consensus (this thread is anything but a consensus), so in that respect, this thread is definitely doing more harm than good IMO.
Post Reply