Problem with Rich Mod (PPIMV) [Archive]

Amit

12-30-2005, 11:46 AM

I have an old NMV PTP amp that is a Marshall Bass clone. I wanted to add a PPI master volume and read good reviews on this Rich Mod:

http://home.houston.rr.com/texast/images/rich_mod.pdf

Found the place where to put the pots (between the caps and 220k resisteors), lifted the caps legs and wired the pots. the issue is that the 220k resistors in the schematic are connected to each other on a common point, but as can be seen in the picture bellow, in my amp they don't connect. a friend told me that since my amp has seperate trim pots for each tube is the reason for this.

http://communabuilder.tapuz.co.il/UsersFolders/temp/images/3012200569716.jpg

I asked him where to connect the third lug of the pots, and he said to wire them to ground since "that's how his Master is wired in his Fender". I did that, the amp sounded like it just got a cold or something... and next thing I saw was my power tubes starting to glow:cool:
I knew that wasn't good, and shortly after that the mains fuse went belly up :confused:

Replaced the fuse and the amp is fine :angel
I still want to make this mod to work, so any one has an idea where the third lug should connect?

Cheers.

http://communabuilder.tapuz.co.il/UsersFolders/temp/images/1406200540837.jpg

Shea

12-30-2005, 01:01 PM

Do not solder the cold lugs of the dual-ganged pots together. You would in most amps, which have a single bias voltage supply and in which the 220k resistors come to a V. But that's not your amp.

Instead, connect a lead from the cold lug of each pot to the end of the corresponding 220k resistor where the bias voltage is connected. That's the end opposite from where you lifted the cap. This way, you will be able to keep your separate bias adjustment trimmers.

I don't like how these instructions leave those caps dangling at one end, not to mention the #2 volume pot. I would leave the caps soldered to their respective terminals, and take the 220k resistors off the tag board and mount them on the pots between the cold lug and wiper lug of each pot. Then disconnect the capacitor ends of the leads that run from the caps to the tubes, and move them to the wiper lugs of the pots. You might need to replace those leads with longer ones. This way, everything will be soldered down securely at both ends.

Also, I'd just leave the #2 channel volume where it is and put the master volume pot in place of one of the speaker jacks. Or, if I had to have it on the front panel, I'd replace the #1 channel volume pot with a dual ganged 1 meg pot, and use that for the volume control of both channels. You'd lose the individual volume adjustment for each channel, but it'd be better than having a volume pot flopping around inside the chassis. Or, if I were absolutely sure that I'd never use channel 2, then I'd remove that volume pot and all the leads that run to it, and maybe run a lead from the coupling cap in that channel to ground.

But before you try any of that, just try doing what I said in my second paragraph, and see if that fixes your mv.

Shea

Wakarusa

12-30-2005, 01:02 PM

The problem is that the point where the 200K resistors tie is not at ground, but at C-. That's the negative voltage feed for the fixed bias to the power tube grids. When you connected your MV pot to tie to ground and then turned it to dump signal you:
- Connected the power tube grids to ground, essentially removing the fixed bias and causing the nasty glow.

- Connected the bias supply to ground. May or may not have caused problems, but the bias circuit isn't typically designed to tolerate a lot of current flow (since it normally feeds into the high impedance of the power tube grid circuit).

You'll notice in the RichMod schematic he ties the MV pot to the bias voltage, not to ground. I'm personally not a fan of this method as it puts DC on the MV pot. A better circuit brackets the MV pot with coupling caps - one end blocks the plate voltage from the PI, the other blocks the bias voltage. The two tricks are:
1) Finding space to mount everything (as you've found)
2) Keeping the total capacitance of the grid circuit correct while maintaing frequency response. There's another issue with phase shift and its impact on NFB, but that's a longer story :)

hasserl

12-30-2005, 01:03 PM

The reason your tubes went into meltdown is because you've broken the bias circuit to them. There is a negatvie voltage suplied from the bias circuit riding along on the grid circuits going to each power tube. By connecting the third post of the pots to ground you're starving the tubes of the bias voltage.

To make it work the third lug of the pots needs to tie back to the other side of the 220k resistors. Since your bias circuit is setup with seperate bias trim pots for each tube you will need to run seperate wires from each of the pot sections. In other words, don't tie the two posts together as instructed on the Rich mod instructions. Instead run seperate wires to the ends of the 220k resistors oposite of where you connected the middle legs of the pots. That should restore the bias voltage and you should be alright.

Let us know if that works or not. I don't use this type of PPI-MV when I install one, I use one based on the Marshall 2150 amp. But it's a little more involved and you need to remove the existing coupling caps and replace with four new ones of double the value. I won't go into that here, but if you can't get this one working right I can send you some info on the other method.

Amit

12-30-2005, 01:45 PM

Wakarusa

12-30-2005, 03:10 PM

But the pots seem to take off just about half the power...When fully CCW the amp still has a healthy amount of output.

This is the other problem with having the MV reference the bias supply instead of ground. There's a tail resistor in the bias voltage divider that the AC (music) signal is still working against and that, in this scheme, cannot be eliminated.

Edit: depending on the actual value of the tail resistor with respect to the 220K feeds you may be able to solve the problem by installing a bypass cap around the tail resistor. The down side is that it will bleed off some signal regardless of where the MV is set.

Shea

12-30-2005, 03:13 PM

Nothing exploded, power tubes seems fine:p

So it sounds like you've fixed the most important problem, good.

But the pots seem to take off just about half the power...When fully CCW the amp still has a healthy amount of output.

Interesting. That might have something to do with your bias circuit. Try adding a 10 mfd / 100v cap in between the cold lug of each pot and ground, or put them on the tag board, in between the bias supply end of each of the 220k resistors and ground. Remember to connect the positive ends to ground and the negative end to the bias supply. If you get oscillations -- which can be evidenced by a high-pitched squeal, or tubes replating for no apparent reason, or weird intermodulation distortion -- then take them back out.

I'm using two 500k pots (not ganged).

So you have to adjust these two pots separately to set your master volume level? Once you get the bugs worked out, it could be fun to experiment with setting them at different levels, to introduce some even-order harmonics.

Shea

Amit

12-30-2005, 06:34 PM

You'll notice in the RichMod schematic he ties the MV pot to the bias voltage, not to ground. I'm personally not a fan of this method as it puts DC on the MV pot. A better circuit brackets the MV pot with coupling caps - one end blocks the plate voltage from the PI, the other blocks the bias voltage. The two tricks are:
1) Finding space to mount everything (as you've found)
2) Keeping the total capacitance of the grid circuit correct while maintaing frequency response. There's another issue with phase shift and its impact on NFB, but that's a longer story

Well first of all I want to hear how my amp sounds with a PPI Master vol, if it will sound good, I will look further into the best way to implement it for good:)

Try adding a 10 mfd / 100v cap in between the cold lug of each pot and ground, or put them on the tag board, in between the bias supply end of each of the 220k resistors and ground. Remember to connect the positive ends to ground and the negative end to the bias supply.

Remember?
Hehehe, I didn't know untill now that caps have +/-:D
I'm learning hands on...
Anyway, found a single 10mf/100v and few 10mf/50v. should I use the 100v and a 50v? two 50v?

So you have to adjust these two pots separately to set your master volume level? Once you get the bugs worked out, it could be fun to experiment with setting them at different levels, to introduce some even-order harmonics.

Shea

Went with two pots because that's what I have around here, but yea, I will probably stay with two pots for the tweekability. I already have in the power section an EL34 and 6550 (bought the amp with two 30+ years original EL34s) so yea, I like to experiment with sounds. for lack of knowing how, I set the bias for each tube by ear (and to some extent by eye). btw I will keep both channel volumes, and will probably use a dual concentric pot in the channel two output jack (I jumper both channels and plug into the first channel so one I/O is not in use).

hasserl Shea Wakarusa thanks alot guys:BEER

Shea

12-30-2005, 11:47 PM

Remember?
Hehehe, I didn't know untill now that caps have +/-:D
I'm learning hands on...
Anyway, found a single 10mf/100v and few 10mf/50v. should I use the 100v and a 50v? two 50v?

Electrolytic caps and tantulum caps are polarized, meaning they have a positive lead and a negative one. Other types don't. But the only 10 ufd / 100v caps that will be small enough to fit in there will be polarized ones.

A 50 volt cap might get by, but it might not. And if it blows up, it will spray a mist of corrosive stuff all over everything around it and smell awful.

If you were to try to use two 50-volt caps, you'd have to put them in series, but that will cut the total capacitance in half, to only 5 mfd. And you'd have to take two resistors of equal value (maybe 500k or 1 meg) and put one parallel with each cap to ensure that the voltage across each cap is half of the total. So, IMO, you might as well go out and get a proper 10 uf / 100v one.

Shea

Amit

12-31-2005, 12:40 PM

ITS WORKING ITS WORKING!!!!:BEER

Not a whole lot of gain, but it does sound fine, no fizzy stuff. still need to experiment.
Nice, at least I can start using it with boost/od pedals again. without the MV I could only use distortion pedals since I never reached even a slight degree of breakup when pushing the front end, just more volume (which is great for clean sounds of course:D )

Two things though. first, it still bleeds with the MV fully CCW, nothing like before, but could use a bit more attanuation for those late nights at home. the other thing is the sound, seems it has plenty of low end, nothing I can dial out with the bass/mid EQ...
Is it because my output coupling caps are 0.1uF?

Cheers!

Shea

12-31-2005, 08:10 PM

How did you improve the attenuation? Did you add the caps I suggested? I'm just curious whether that helped, if you tried it.

Yeah, you can cut bass by reducing the size of those two 0.1 caps. A lot of Marshalls use .022 in that position. Save the originals, though.

Shea

Amit

12-31-2005, 10:16 PM

Yup, did it with the 10uF caps between third lug and ground.
Should I go lower/higher to achive more attenuation?
What effect will have if changing the pots to 250k or 1M?

I will sure keep the coupling caps.
Should I look for any specific kind/type of cap for this spot?

Thanks again:)

Scott_F

01-01-2006, 09:41 PM

http://blueguitar.org//new/schem/misc_mod/pi-m-vol.pdf

donnyjaguar

01-02-2006, 12:21 AM

I think there is more merit putting a variable gain control outside of the negative feedback loop myself.

Slick51

01-02-2006, 01:09 AM

http://blueguitar.org//new/schem/misc_mod/pi-m-vol.pdf

Scott

This schematic won't work because of the separate bias pots Amit mentions... this wiring results in the ground vs. bias reference problem that Todd, Shea and Richard describe above.

Slick51

Wakarusa

01-02-2006, 09:53 AM

Scott

This schematic won't work because of the separate bias pots Amit mentions... this wiring results in the ground vs. bias reference problem that Todd, Shea and Richard describe above.

Slick51

No, don't think so. The MV pots (which do, IMHO, work better as a ganged set, though a concentric set might be cool) are DC isolated by 0.1uF caps on either side and referenced to ground. The two items I would point out are:

- Using 0.1uF coupling caps yields an effective capacitance of 0.05uF with attendant frequency response changes and an additional phase shift - something to consider with respect to the phase of the NFB signal.

- The 33-100K resistors from MV wiper to grid coupling cap aren't strictly required. I'm guessing that the designer didn't want the power tube grids tied directly to ground, though I'd have to think a bit to figure out why. They have an impact on frequency response too.

Wakarusa

01-02-2006, 10:02 AM

I think there is more merit putting a variable gain control outside of the negative feedback loop myself.

Lots of designs throw a pot in before the PI input coupling cap (or earlier). Fewer/cheaper parts is one of the advantages. This gets into the long standing debate over whether a MV should go pre or post PI. I think the best answer ultimately depends on whether there's enough gain on tap in the preamp to push the PI into compression/distortion. In the case of a long-tailed pair, as the input drive goes up typically the imbalance between the outputs increases too (even before compression/clipping) producing a 2nd order harmonic distortion that some folks like. Putting the MV before the PI removes this stuff from your "low power" tone.

I suppose in a perfect world, you'd put MV controls both before and after the PI (and perhaps at each gain stage) for ultimate tweakability, but you quickly run out of front panel real-estate and just as quickly bump up against how many knobs a player is willing to fool with.

Shea

01-02-2006, 10:13 AM

Yup, did it with the 10uF caps between third lug and ground.
Should I go lower/higher to achive more attenuation?

Definitely not lower.

Whether a bigger cap would help depends on what it sounds like when your two mv pots are turned all the way down. If the tone is particularly bassy, with no treble, then using bigger caps will help.

If the tone is all treble, no midrange or bass, then try bypassing those 10uf caps with some smaller ceramic caps, maybe .01 or .001 in value.

But if the tone sounds fairly normal, even with the mv pots turned all the way down, then I think the problem is where you grounded those caps. In that case, you'd want to experiment with grounding them at the same place where the other bias supply filters are grounded, or where the power tubes' cathodes are grounded, or where the first filter cap is grounded, or where the tail resistor of the phase inverter (or the presence pot) is grounded.

What effect will have if changing the pots to 250k or 1M?

Different taper, possibly different bass response. You might get more bass with 1M pots when they're turned down, but when you turn them all the way up, the bass would drop back down to about what it is now.

If you replaced the pots with 250k's, then you could increase the value of fixed resistors that are soldered between the cold lugs and wipers to 500k or maybe 1 meg. Then the pots will account for most of the loading on the phase inverter, and the fixed resistors will just be there to ensure that the tubes still have fixed connection to the bias voltage supply if the wipers of the pots make poor connection (such as if the pots get dirty).

Another way to do it is to use 250k linear taper pots, switch the leads that are connected to the wipers and the hot lugs of the pots, and get rid of the fixed resistors entirely. So, you would have the outputs from the phase inverter connected to the wipers, and the leads that run to the grids of the power tubes will be connected to the hot lugs of the pots. It's basically wiring the pots backwards. You'd still connect the cold lugs to the bias supply. This way, the pots completely replace the 220k resistors that are in the original, non-mv circuit. And by wiring the pots backwards, the power tubes won't lose their bias voltage if the pots get dirty. This does work - I used an mv like this in a silverface Twin Reverb for a while. But eventually that amp quit working, for reasons that I never bothered to figure out. Maybe it had to do with the pots failing, or maybe not. So use at your own risk.

Shea

Slick51

01-02-2006, 12:33 PM

No, don't think so. The MV pots (which do, IMHO, work better as a ganged set, though a concentric set might be cool) are DC isolated by 0.1uF caps on either side and referenced to ground. The two items I would point out are:

- Using 0.1uF coupling caps yields an effective capacitance of 0.05uF with attendant frequency response changes and an additional phase shift - something to consider with respect to the phase of the NFB signal.

- The 33-100K resistors from MV wiper to grid coupling cap aren't strictly required. I'm guessing that the designer didn't want the power tube grids tied directly to ground, though I'd have to think a bit to figure out why. They have an impact on frequency response too.

Not referring to the voltages on the MV pot; referring to the grounded grids. The problem was that he'd wired the amp as though it had the shared bias/both 220ks tied together, but he had shunted the bias to ground (as you describe in post 3; that's why I specifically referered to you in the reply). The blue guitar schematic shows the shared bias situation, not the separate bias pots/220ks inline that he had.

I just didn't want to muddy his repair effort by redirecting him back to a schematic that doesn't apply directly to his amp. Not to be disagreeable; just trying to help.

I use the series caps to keep both the B+ and C- off the pot as well, and if I have plenty of gain, the series resistor is a good/safe idea too. I can deal with the phase/freq shift more than I can the scratchy pots.

Slick51

Wakarusa

01-02-2006, 01:04 PM

Makes sense.. didn't understand your original intent.

Agreed about keeping DC off the pots being paramount. For others watching, the phase shift issue isn't rocket science (heh.. I've done rocket science, so I know the difference ;)). Just requires some careful math and measurement and, possibly, some reactance in the NFB line.

Shea

01-02-2006, 01:53 PM

Agreed about keeping DC off the pots being paramount.

What makes it paramount?

Shea

Slick51

01-02-2006, 01:57 PM

No problemo. I'm often misunderstood, even by me!

What rocket science did you/do you do? I'm a "Certified Space Professional" by the USAF (a no-joke certification, regardless of job speciality code/education). Nice to meet another space geek. Cool beans.

Slick51

Slick51

01-02-2006, 01:58 PM

What makes it paramount?

Shea

To me it's a 'relative paramount'; more important than minimizing freq deltas and phase shift. Scratchy pots belong on the ZVex SHO!:D

Slick51

Wakarusa

01-02-2006, 02:55 PM

Slick51

01-03-2006, 09:40 PM

Ahhh...Kepler doesn't like me. Every time I get close to him he speeds up.:D BTW, I used to work a couple of doors down from the GPS controllers, in the late 80's when we were getting it all set up.

On those fixed resistors after the caps; Shea mentioned the ways that Kevin O'Connor described in "The Ultimate Tone" to get rid of one pair of caps, including the reverse method. I remembered reading that too, so I pulled TUT off the shelf and reread it. Seems like Kevin was more concerned with the high pass filtering of the signal at low volumes due to the low resistance the pot provides to the RC network. By putting that resistor in series, you limit the upper travel of the low-frequency rolloff (as Todd alluded).

In a typical circuit with a 0.022uf coupling cap and a 500k log pot on '1', the rolloff frequency is 1045Hz, a severe bass loss. With the 100K resistor inline, the rolloff starts at 72 Hz, worst case.

Another thing arising from the filtering is the effect of sequential RC networks on the cutoff frequency and slope. The first filter knocks off bass at 6dB/octave (1st order); the second at 12dB/oct (2nd order), etc. and the frequency creeps up too. In a JCM800 circuit, the third filter's rolloff is 720Hz, at 18dB/oct.

Mr O'Connor suggests using the pair of caps around the MV pot, using the original caps and a second set at least 10 times larger than the first set to effectively take the rolloff out of the equation. I did the math, and the difference in total capacitance for a 0.047uf cap/0.47uf cap pair is 0.004 (0.043uf total), insignificant relative to the tolerance value on the cap. Plus, you don't have to replace the first cap, and the second cap only has to handle C- voltage levels (a 100 volt cap is plenty).

The only thing I haven’t quantified is the signal loss due to the inline resistance, but I think it would be less noticeable than the bass rolloff. At low volume, you don’t care about a little more loss; at high volume you have plenty of gain.

Glad I did the reading.

Slick51

hasserl

01-04-2006, 09:44 AM

Hey Slcik, that's pretty cool! I may use that, thanks.

Hasse

Shea

01-04-2006, 10:18 AM

On those fixed resistors after the caps; Shea mentioned the ways that Kevin O'Connor described in "The Ultimate Tone" to get rid of one pair of caps, including the reverse method. I remembered reading that too, so I pulled TUT off the shelf and reread it. Seems like Kevin was more concerned with the high pass filtering of the signal at low volumes due to the low resistance the pot provides to the RC network. By putting that resistor in series, you limit the upper travel of the low-frequency rolloff (as Todd alluded).

In a typical circuit with a 0.022uf coupling cap and a 500k log pot on '1', the rolloff frequency is 1045Hz, a severe bass loss. With the 100K resistor inline, the rolloff starts at 72 Hz, worst case.

I used that type of mv in a 70s Twin Reverb, which had .1 blocking caps instead of .022. Using those caps, I never noticed a problem with excessive bass rolloff. Maybe there was at bedroom volumes, but I never played it that low.

I added some brightness caps, and I thought the value I used was pretty good at making the high end sound about the same at lower levels. It was probably 120 pf, because I had some of those laying around.

I could have sworn that this two-blocking cap type of mv was more clear and transparent sounding than the 4-cap type that O'Connor ended up advocating. Maybe it was just my imagination (you know, that audio placebo effect that applies to mods and component upgrades), but I really did think so.

Another thing arising from the filtering is the effect of sequential RC networks on the cutoff frequency and slope. The first filter knocks off bass at 6dB/octave (1st order); the second at 12dB/oct (2nd order), etc. and the frequency creeps up too. In a JCM800 circuit, the third filter's rolloff is 720Hz, at 18dB/oct.

I think the steep slope of the bass rolloff in Marshalls is probably the key to how they're able to pass through just enough bass to sound big and mean without ever getting flubby. (Of course, the cab helps too). That "slim and smooth" pedal probably works on the same principle - sequential high pass filtering.

Shea

Slick51

01-04-2006, 05:27 PM

Shea

I think you've nailed several of the apparent phenomena going on. The worst case bass loss is at bedroom volumes; by MV position = '5', the rolloff is below the low E fundamental. However, when people install MVs, it is often to get that good tone at MVs around '3'. The loss is also accentuated by the Fletcher-Munson curve too...

The end result is always, 'how does it sound to me'? If I'm not doing an A/B test, I usually just adjust it to where I like, and stop. That's why we have EQs, etc.

I agree the Marshall sound is achieved largely by the low bass trimming before/during the preamp chain. The Marshall example I used was just to show what frequencies are affected. If one notices that the low E string doesn't thump after the PPIMV mod, this is why. Then, if one wants to address it, there's where to stop.

I've read about the Smooth and Slim, but never heard/seen one. Do you oen one or have experience with it? Just wondering...

Finally, the reverse pot method solves that bias loss if the pot fails. I've used that method on a few amps too.

Great hobby, and site. I'm continually amazed by the knowledge and attitude here. Thanks for the input, EVERYONE.

Slick51

Shea

01-04-2006, 05:48 PM

I've read about the Smooth and Slim, but never heard/seen one. Do you oen one or have experience with it? Just wondering...

Nope, but I'd really like to have something like that, to make the bass knob on my Fenders more usable with overdriven tones. I've tried putting in smaller cathode bypass caps and blocking caps, but that way you lose the big Fender bass for clean tones. And what I ended up with sounded too Boogie-ish.

If I ever find the time, I'll try making my own low-end filter pedal. I figure a unity buffer, followed by maybe 3 highpass filters, with a rotary switch to select whether the output is taken after the 1st, 2nd, or 3rd filter. The input to the buffer could be designed to cut out a little low end too.

Shea

Wakarusa

01-04-2006, 06:21 PM

Probably pointing out (or restating) the obvious, but the bass rolloff from the 4 cap PPIMV is often desireable. Bedroom levels and clean with full range response is max'd MV, adjust channel volume to taste. For grind channel volume goes up, MV comes down, often with the attenuation of bass that can be perceived as a boost in treble giving exactly the driven tone folks have come to expect.

Related story -- there's a well-known rolloff in treble response in your average guitar volume circuit as volume is reduced. So if you set your amp to play clean at reduced volume and then use the guitar volume to swell into distortion/compression you get exactly the effect described above (though in this case it is a boost in treble response instead of attenuation of bass). This also implies that the same solutions to mitigate the treble rolloff on the guitar volume will work on the PPIMV circuit ;)