JTM45 Phase Inverter Mods

[sgbrewing]

Hi,  I built a jtm45 nearly a year ago now and it sounds great! I use a PPIMV Rich Mod II for practice at home to not annoy the mrs or baby :D
The sound is very good at low volume, best I've gotten yet but I'm looking on a couple tweaks to fine tune it.

As the PPIMV is down low the PI tube is getting slammed pretty hard so I'm thinking of trying 470K grid stoppers on the PI socket to help smooth out the gain a touch. Would I use two grid stoppers on both pin 2 & 7? or just one side?

I'm also thinking of reducing the PI input cap from 0.022 to either 0.01, 0.0047 or 0.0022 to help with the blocking distortion cause by slamming the PI, how would these values change the tone?

The other changes I've made from the stock circuit are:
470k mixers
5uf v1 cathod cap
and PI coupling caps reduced to 0.047uf

Cheers!

[66Strat]

I thought that slamming the PI was the whole point of a PPIMV.

[sgbrewing]

I thought that slamming the PI was the whole point of a PPIMV.

Yea of course, but I just want to smooth out the gain / reduce some blocking distortion.

[66Strat]

Try grid stoppers on the power tubes in the 30K to 50K range.

[sgbrewing]

Try grid stoppers on the power tubes in the 30K to 50K range.

But the power tubes aren't getting pushing with the PPIMV really low, how would this help?

[SnickSound]

I thought that slamming the PI was the whole point of a PPIMV.

Yea of course, but I just want to smooth out the gain / reduce some blocking distortion.

That's why... unpopular opinion time... I prefer to have both a pre and post MV.

But yeah, series resistance between the tone stack and PI input would work. No need to add grid stoppers on the PI tube, just put a resistor right on the output of the Treble cap (assuming no FX Loop). Start at 220k.

You could even hardwire a set pre-PI MV by having a resistor on the wiper than another to ground, take the signal at the junction. This would reduce drive to the PI.

Of course, many will say all this nastiness you get from overdriving the PI is part of the JTM45 sound. But it's your amp and experimenting is fun.

[66Strat]

The grid stoppers inhibit the ability of the power tubes to draw grid current. The grid current builds up a DC charge on the coupling cap that causes blocking distortion. It's not just a matter of signal amplitude. Frequency and cap size also come in to play.

[sgbrewing]

I thought that slamming the PI was the whole point of a PPIMV.

Yea of course, but I just want to smooth out the gain / reduce some blocking distortion.

That's why... unpopular opinion time... I prefer to have both a pre and post MV.

But yeah, series resistance between the tone stack and PI input would work. No need to add grid stoppers on the PI tube, just put a resistor right on the output of the Treble cap (assuming no FX Loop). Start at 220k.

You could even hardwire a set pre-PI MV by having a resistor on the wiper than another to ground, take the signal at the junction. This would reduce drive to the PI.

Of course, many will say all this nastiness you get from overdriving the PI is part of the JTM45 sound. But it's your amp and experimenting is fun.

Thanks for your reply!

To be honest the PPIMV with this amp actually sounds really good low. I'm not really looking to reduce the gain, more just tiny adjustments just to smooth it out a touch.

If I was to add grid stoppers at the PI tube would I need 2, 1 on each side of the tube? Or just 1 side?

Cheers!

[SnickSound]

If I was to add grid stoppers at the PI tube would I need 2, 1 on each side of the tube? Or just 1 side?

Cheers!

Not that I have ever tried that, but it would make sense that only one side would be required (the one where the PI input coupling cap is fed into)
The 2nd side is where NFB is fed, and that actually reduces distortion (although it becomes kind of irrelevant when the PPIMV is turned down, the signal is too low to make a difference). There wouldn't be any benefits in grid stopping that.

[2deaf]

If I was to add grid stoppers at the PI tube would I need 2, 1 on each side of the tube? Or just 1 side?

Both grids on the PI can draw current and both are connected to capacitors, so both can cause blocking distortion.  Lowering the value of each cap and installing grid stoppers on both grids helps.

[2deaf]

As the PPIMV is down low the PI tube is getting slammed pretty hard so I'm thinking of trying 470K grid stoppers on the PI socket to help smooth out the gain a touch.

470K seems a little large for a stage with Miller capacitance unless you want to shear off some high frequencies. 

[sgbrewing]

As the PPIMV is down low the PI tube is getting slammed pretty hard so I'm thinking of trying 470K grid stoppers on the PI socket to help smooth out the gain a touch.

470K seems a little large for a stage with Miller capacitance unless you want to shear off some high frequencies.

Lovely thank you, I have some 470K arriving soon so I'll try the first side with the 0.022 and drop that cap and see how it effects the sound and then add one to the other side.

Yea the PPIMV adds some fizz so hopefully these changes will reduce it :)

Thanks!

[tubenit]

Do a search for "enhance cap" and read the thread, please.

My suggestion before doing anything else is to try a 120p -220p range cap across the plate resistor of the signal side of the LTPI.  It smooths the highs very nicely in my experience.

With respect, Tubenit

[sgbrewing]

Do a search for "enhance cap" and read the thread, please.

My suggestion before doing anything else is to try a 120p -220p range cap across the plate resistor of the signal side of the LTPI.  It smooths the highs very nicely in my experience.

With respect, Tubenit

Cheers that looks great! I'll definitely try that, will order a couple values of a few different types of caps :)

Shall I keep the 47pf fizz cap with this?

So I attach your cap to the plate resistor that' connected to the PI input cap?

My mixer bright cap is a 500pf silver mica, i've heard that a standard ceramic might be a touch smoother also?

Thanks!

[tubenit]

Shall I keep the 47pf fizz cap with this?

I don't know?  Your call on that.  Try with it in first.

So I attach your cap to the plate resistor that' connected to the PI input cap?

There is a schematic with this thread    https://el34world.com/Forum/index.php?topic=12723.msg153636#msg153636

My mixer bright cap is a 500pf silver mica, i've heard that a standard ceramic might be a touch smoother also?

I have found some brands of ceramic caps in that value to be smoother than silver micas.

With respect, Tubenit

[pdf64]

If I was to add grid stoppers at the PI tube would I need 2, 1 on each side of the tube? Or just 1 side?

Both grids on the PI can draw current and both are connected to capacitors, so both can cause blocking distortion.  Lowering the value of each cap and installing grid stoppers on both grids helps.

For the non inverting (feedback) input, I can see how reducing its cap value would help with bias shift / blocking distortion for the given scenario. But can’t see how a grid stopper would help there? ie as it’s decoupled to (near enough) circuit common, how can the grid’s potential get pulled up above its cathode, (which would be necessary for grid current to flow and hence a grid stopper be beneficial)?

[2deaf]

For the non inverting (feedback) input, I can see how reducing its cap value would help with bias shift / blocking distortion for the given scenario. But can’t see how a grid stopper would help there? ie as it’s decoupled to (near enough) circuit common, how can the grid’s potential get pulled up above its cathode, (which would be necessary for grid current to flow and hence a grid stopper be beneficial)?

Aren't you acknowledging grid current when you agree that reducing its cap value would help with bias shift/blocking distortion?  Where does the current come from that charges this cap for grid shift/blocking distortion?  You don't think that the connected cathodes can swing far enough negative to make the non-inverting grid-to-cathode voltage positive?   

[pdf64]

For the non inverting (feedback) input, I can see how reducing its cap value would help with bias shift / blocking distortion for the given scenario. But can’t see how a grid stopper would help there? ie as it’s decoupled to (near enough) circuit common, how can the grid’s potential get pulled up above its cathode, (which would be necessary for grid current to flow and hence a grid stopper be beneficial)?

Aren't you acknowledging grid current when you agree that reducing its cap value would help with bias shift/blocking distortion?  Where does the current come from that charges this cap for grid shift/blocking distortion?...

My thinking is that when overdriven, the dc operating point may shift around a bit. Long time constants at the grid will tend to slow down its ability to keep up with those changes.

...You don't think that the connected cathodes can swing far enough negative to make the non-inverting grid-to-cathode voltage positive?

Haha, I'm having a hard time getting my head around it, with that triode operating in common grid mode and the dc elevation.
It would be nice to get it on the scope, but as I don't have a galvanic isolation transformer for it, I suspect that without that, the scope inputs would load the circuit too much.
The LTP cathode of an overdriven in a JTM45 type circuit can certainly respond to the signal, Vp-p there looks about 5V. I think that if I add a grid stopper to the non-inverting section's grid, eg 10k, grid current would manifest as VDC appearing across it, when the signal exceeds a certain level. Does that sound right? If so, I'll try to check it out.

[SILVERGUN]

As the PPIMV is down low the PI tube is getting slammed pretty hard so I'm thinking of trying..........................

Have you tried replacing the PI tube with either a 12AT7 or even a 12AU7?
I know what you're thinking, but you won't know until you know. I like the AU7 for it's ability to handle a much larger input signal.
If you really want to experiment, I would suggest halving the plate resistors and doubling the cathode resistor value to get the AU7 closer to the sweet spot.
I obviously have no loyalty to the original circuit.

[2deaf]

The LTP cathode of an overdriven in a JTM45 type circuit can certainly respond to the signal, Vp-p there looks about 5V. I think that if I add a grid stopper to the non-inverting section's grid, eg 10k, grid current would manifest as VDC appearing across it, when the signal exceeds a certain level. Does that sound right? If so, I'll try to check it out.

Given the magnitude of the grid current, I'd go with 470K to make the detection of VDC easier.

[shooter]

lurking here;

Given the magnitude of the grid current

does blocking distortion favor fixed or self bias 

[2deaf]

does blocking distortion favor fixed or self bias

I don't know.  The only time I ever fooled with a fixed bias LTP was when the inverting grid was connected directly to the plate of the preceding stage.  There was no capacitor there to create blocking distortion.  The non-inverting grid could still create blocking distortion, but it was a Hi Fi attempt and overdriving wasn't allowed.

[DummyLoad]

has anyone tried using a CCS for the LTP tail R?


--pete

[pdf64]

The LTP cathode of an overdriven in a JTM45 type circuit can certainly respond to the signal, Vp-p there looks about 5V. I think that if I add a grid stopper to the non-inverting section's grid, eg 10k, grid current would manifest as VDC appearing across it, when the signal exceeds a certain level. Does that sound right? If so, I'll try to check it out.

Given the magnitude of the grid current, I'd go with 470K to make the detection of VDC easier.

I added a 100k grid stopper to the LTP's non-inverting input of a JTM45/50 kinda thing that's on my bench, it already had a 100k grid stopper on its regular inverting input. A Type 3 master vol was connected and set to approx. halve the VAC at the speaker output, resistive load. The presence control was set to minimum.
Even with the LTP heavily overdriven by a 1kHz sine at the normal input, I couldn't measure anything >1uVDC across either 100k LTP grid stopper (Fluke 189).
Attempting to connect a 10:1 scope probes there affected the output waveform a little, unfortunately.
Connecting a shorting link across the inverting input grid stopper affected the waveform a little, whereas connecting it across the non-inverting input's grid stopper had no effect.
Tests were repeated with presence control at max but the results were the same.

[2deaf]

No power tubes.  No presence.  220K grid leaks.
R1 is the grid stopper for the inverting stage.
Out 1 is the plate signal of the inverting stage.
R2 is the grid stopper for the non-inverting stage.
Out 2 is the plate signal for the non-inverting stage.

R1 alone has a significant effect on grid-current clipping at both Out 1 and Out 2.  The negative side of Out 1 is clipped and the positive side of Out 2 is clipped by grid current.

R2 alone only has a minor effect on both outputs.

With both R1 and R2 installed, the grid-clipping effect is slightly more than with just R1 alone.

 

[2deaf]

Both stages have a Miller Effect.

With R1 alone, the high frequencies are attenuated the same on both outputs.

With R2 alone, the high frequencies are attenuated about the same on both outputs, but the frequency response curve is shifted to higher frequencies as compared to R1 alone.

With R1 and R2 installed, the high frequency attenuation is like a second-order filter with its steeper slope.