V1, V3, V4, V5, V6. Tung Sol 12AX7 in V2 will make the edge mode on CH2 sound really good. Mullard long plate (current production) 12AX7 in V7 for the phase inverter. Some like the Sovtek LPS but the Mullard is the same tube but of better quality. Mesa stock is ok there as well. Best to have matched triodes for the phase inverter. The Mullard enhances the low end and top end with a good balance. The phase inverter tube is important and will influence tone.
What I ran in my Mark V90 was a bit different.
V1 and V3 were Svetlana 12AX7 (current production tubes, similar to the Mullard CV4004) They eliminated the 60Hz hum I was getting on the clean channel, it was so annoying, now it is completely gone. V2 I have a Tung Sol 12AX7, V4, V5, V6 were the Chinese Bejing tubes and V7 was a Mullard 12AX7 (current production, reissue long plate). I have had many issues with the Mark V90, so far I have yet to have the diodes fail. It is a 2012 so that particular year seemed to be prone with component problems. I do not favor the Mark V90 much. Not a bad amp but mine needs something fixed in it go correct the ice pick problem, I believe it may have more to do with the GEQ circuit than anything, it is not Just CH3, CH1 tweed, Ch2 edge were nasty ice pick makers too but the change in preamp tubes did help with that. I now like the edge mode, used to hate it. I still have an issue with CH3 though. I put the amp away out of sight and will get back to it later when I am ready to remove the main board to fix it. First, I will need to run test on a live preamp with an oscilloscope to see if I can locate the issue. Waveforms from the send jack were not correct, will have to check each gain stage to see if it is from that or is it from the GEQ circuit (which I suspect). Something is cutting the lower frequencies in the upper swing of the waveform. This is just a single frequency at 750Hz being injected into a Lehel P-split II with the isolated output used to drive the amp. This was to keep any DC voltages from passing into the front end. Also, did not want to find out what will happen to the function generator if there were any potential ground loop problems back feeding into the generator from the tube amp. That could damage both amp and generator. The Tera Echo was just a load on the FX loop, in a hard bypass mode.
View attachment 4624
It should look more like this (from the JP2C)
View attachment 4625
Originally , I was only doing a test to determine the signal levels from a few amps. At least I have some idea what may be going on with the V. Those peaks and drop out of the lower frequency on the upper portion of the waveform will have an ice pick tone. the bottom portion has more curvatures to it than the JP2C, that indicates some reduction in higher order harmonics. The flatter the curve, the higher the harmonic content is. Typical frequency harmonic range of the Mark III when I had the opportunity to run a signal analysis on it was 4th order before they dropped off. The order is a multiple of the original frequency. Even order harmonics will sound pleasing if you can get some roll off before you reach the 8th order or 8 times the fundamental frequency. Odd order harmonics sound harsh except for the 1st order which is only 1 x fundamental frequency. 5, 7, 9 will make things sound brittle. Note the sharp leading edge and the flat line on the upper waveform, it is not as sharp as it looks and the corners are well rounded on a close up view. That is with all of the JP2C controls set to noon. Not using the GEQ either but that is after the FX loop so its influence would not show up in the plots. The Mark V90 on the other hand uses the GEQ signal to create the FX send level so it will always be in the circuit even when turned off. Also it will be in the signal path in the FX loop hard bypass mode. so no way to avoid it if that is the root cause of the ice pick.