Univox U255r- what is this phase inverter?

[imaradiostar]

In light of the other thread on a univox I've been inspired to post a thread about mine.

I've got a U255R on my bench that I'm working on for a friend. I'm not familiar with the phase inverter design so I thought I'd ask. I'm confused how it's supposed to work and can't help but think that with a few small mods to a split load style it'd sound much better.

here's a link: http://www.univox.org/pics/schematics/u255r.gif

The amp sounds pretty miserable as-is and I was considering carefully removing the stock guts and putting them in a baggy just in case they ever become desirable. Then I'd build a nice little turret or eyelet board and build something that'll fit the same mounting points but be a little more my style- maybe with a 6ea8 or 6u8 in place of one of the 12ax7's and a solid-state trem oscillator.

Anyone have a compelling reason why I should stick with the stock circuit?

jamie

[Lauri]

That is a split-load phase splitter. The input resistance of the phase splitter is used as a load for the previous stage.

[imaradiostar]

thanks!

what's the benefit of this connection versus the normal way we see them?

jamie

[Jazz P Bass]

Split load PI

The benefit may be that it uses 1/2 of a tube.
Read this:Tales From The Tone Lounge:Mods and Odds!

[imaradiostar]

OK, maybe I'm missing the point here...

I can see that it's a split load inverter but it has connections that make no sense to me.

1)why is there a 5u cap from the cathode of the second stage to the cathode of the inverter?

2)why is the PI plate load so different from the cathode load?

3)why would they waste all that voltage with the 2 100k resistors forming a voltage divider at the plate of the second stage?

4)why would they capacitively couple the second stage plate dropping resistor to the phase inverter cathode?

I understand how a normal split load phase inverter works in both cathode and fixed bias configurations. It would appear that they've created a very strange fixed bias arrangement but it still doesn't explain the differences in split load resistor values.

jamie

[Lauri]

OK, maybe I'm missing the point here...

I can see that it's a split load inverter but it has connections that make no sense to me.

1)why is there a 5u cap from the cathode of the second stage to the cathode of the inverter?

I think that might be a mistake in the schematic

2)why is the PI plate load so different from the cathode load?

Because that cathode load 100k resistor is in parallel with the 100k resistor from power supply.

3)why would they waste all that voltage with the 2 100k resistors forming a voltage divider at the plate of the second stage?

4)why would they capacitively couple the second stage plate dropping resistor to the phase inverter cathode?

The idea behind this circuit is that the second 12ax7 stage sees the phase splitters input resistance as it's load. That way the gain of that second 12ax7 stage is somewhere around 80 even without bypassing the cathode resistor.

There is a example of this circuit in Radiotron Designer's Handbook, fourth edition
Chapter 12 Audio Frequency Voltage Amplifiers, 6: Methods of Exciting Push-Pull Amplifiers. http://pmillett.com/Books/RDH4.pdf

[imaradiostar]

OK, that makes more sense.

Wouldn't the input impedance of the phase inverter still be very high because of the split load inverter without any of the applied shenanigans- in other words I don't understand why it isn't wired like a fender princeton reverb.

The tone circuit is responsible for about 20dB loss...kinda ridiculous on such a simple amp. I think a better tone circuit and more overall treble boost would be a huge improvement.

jamie

[imaradiostar]

I read that portion of RDH4 and worked through the math- it makes much more sense now.

Thanks for posting the example, that's exactly what I was looking for!

Of course... it still sounds terrible! Like I said, I think a lot could be gained by using a better tone circuit and a more normal phase inverter layout.

jamie

[Lauri]

There is a more detailed description of that phase splitter circuit in this Wireless World article.
http://vuojolahti.com/~papu/putki/pd..._WWAug1947.pdf

[Enzo]

It may sound terrible, but are you sure it is the contribution of the PI making it so?

[imaradiostar]

It may sound terrible, but are you sure it is the contribution of the PI making it so?

Nope, not sure. I know the amp doesn't sound good and I'm going to try some other things first. I know many of the caps on old univox amps can be leaky but I've spent some time with a VTVM and I haven't found any that are obviously bad. I'll be getting a digikey order in the mail tomorrow so I'll probably replace any of the questionable caps before I resort to drastic changes.

Have you ever heard one of these things before?

It's a decent test bed for different ideas so I might just try a few different things for fun. I like the simplicity of the circuit.

jamie

[imaradiostar]

I decided to give this thing a chance...so I removed all the electrolytic and oil/paper caps and replaced them with modern components of the same or similar values. It sounded pretty good but the tone control was pretty useless- it mostly sounded like a slight bass roll off while lacking needed treble.

I ripped out the Univox tone circuit and wired up a Ghia/Gibson style control. It's quite a bit more useful- it allows a nice mid scoop in the middle and varying amounts of treble or bass roll off at the extremes.

I added a grid stopper on the phase inverter but there is still some blocking distortion (less than there was) and some PI imbalance at higher drive levels. It sounds pretty good at low to moderate settings and should work well with pedals so it's a big upgrade from where it was. It's not really a "crunch" amp so I'm not concerned that it gets a little strange at higher (more distorted) volumes. You can't have a mid scoop after the first gain stage and expect an amp to crunch like a marshall! I thought about moving the tone behind the second gain stage but that would put the tone in the feedback loop and I didn't want to do that. If I was going to I'd convert it to EL84's and lose the NFB anyway.

I'm still tempted to mod the PI but I don't think it's really worth it. It's usable now and sounds pretty decent so I'm gonna let it be.

The cabinet is really entertaining- it's a waist-high 2x12. It's different from the usual FMV and very usable for a small gig or practice.

jamie

[pbjbryan]

Sorry for reviving an old thread, but I recently acquired a U-255r myself, and spent some time puzzling through the same issues as the OP.

First, the tone control is messed-up, but the fix is very easy. First, reverse the leads to the top and bottom of the tone pot, then move the end of the 500p cap from the wiper to the now top leg of the tone pot (where the 220k goes to ground), and it will work just like the classic Valco/BMP tone circuit. Also adding a 47p across the top and wiper of the volume pot helps maintain brightness at lower volume.

The PI is certainly clever, but in that "fine line between clever and stupid" sort of clever. I.e., on the other side of clever. I personally don't like the sound of overdriven cathodyne (split load) PIs because the "tit" on the waveform caused by grid conduction of the output tubes sounds terrible. Grid stoppers help, and my amp came with 10k grid stoppers that aren't on the schematic. The PI can sound more normal in overdrive by removing all the cleverness. Just remove the 5u to the plate resistors, put a 100k CC in the PIs plate, and take the end of the 5u connected to the non-inverting output to ground (use it as a cathode bypass on the 2nd gain stage).

Another useful and simple change is to rewire the first power cap directly to the rectifiers and power transformer to get the filter return current off the PCB ground. Changing to a 33u/450V cap is recommended since the original 20u/350V cap is under spec'ed. If you like the sag of tube rectifiers, add a 150r 5W power resistor between the power rectifiers and the filter cap.

For even more hum reduction, beef-up the negative bias supply cap to a 33u/50V or bigger. It won't affect the behavior great bias-trem circuit at all.

The reverb return circuit literally sucks the tone out of the signal at high reverb levels, but this can easily be rewired as a reverb balance control by grounding the wiper of the pot, and feeding the ends with the dry (from the 220k) and reverb (from the 70k) signals, and feeding the grid from the ends through two 470k resistors. This lets you go from completely dry to completely drowning in reverb. Very cool for that groovy 60's era sound.

Also, the reverb circuit grounding leaves a lot to be desired, but is also easily fixed. Just remove the ground wire to the PCB, snip the ground between the RCA jacks, and run separate grounds for each jack. Wire the return jack ground back to the input ground, and run the send jack ground to the bottom of the 150r driver stage cathode resistor. While you're at it, just completely remove the return trim pot and replace it with a 47k metal film resistor grounded to the return jack's ground lug. It's unnecessary, and the 50k pot can be used for better things.

Forgetting all the blather about cathodyne or long-tail pair PIs, I personally like the very simple cascade phase inverter like the ones use in early Fender tweeds and most of the Valco amps. This is a decidedly non hi-fi circuit and rightly so. These are GUITAR amps, not stereos, and they should have some musical distortion fer crissakes! The cascade PI adds some nice 2nd order harmonic to the output, which is very musical and helps improve clarity and detail.

The repurposed B50k trim pot from the reverb return circuit makes a perfect gain balance pot for the 2nd stage of the cascade inverter. Just rewire the PI as a normal gain stage, and couple it to the 2nd gain stage through a big resistor feeding the 50k trim, and adjust for unity-ish gain.

Here's a doctored photo of pcb wiring side showing the top side components. And another of the pcb with the part values drawn in.

[imaradiostar]

Lots of interesting info here! I chose to stick with minimal mods on the one I had at the request of the owner. It sounds like you really made a neat amp out of it! If I happen upon another one some day I'll have to try some of your ideas.

I think your post reads like a primer of common designs problems and fixes. There are many amps that have noise issues or performance issues that could be sorted out with a few minor changes. It's neat to read how thorough you were in re-engineering the amp!

The speakers in the one I had were awful, just really bad. How were yours?

jamie

[pbjbryan]

The speakers in the one I had were awful, just really bad. How were yours?

Not bad sounding, but not efficient at all. They're black-frame Jensen Special Design label speakers, both date code 220931 (July 1969), and marked C8171-6. There are no other markings, but based on similar models, these are an OEM variant of the P-12S modified to have large aluminum dust caps. The small alnico magnets and 1" voice coil don't put out very much sound.