Correcting the Marshall Tremolo Circuit

I needed to get the Marshall as good as I can for my level of amp repair knowledge, as it has been bugging me for ages as a loose end since the main “just get sound out of it” repair was done months back – the Marshall Repair Posts.
It became even more of an issue for me after I saw my lecturer, Dave’s 1972 Marshall 25W all valve with Tremolo model, in the flesh last week, but he hasn’t seen this one yet.
His tremolo is fine, and sounds great – a classic vintage sound. As I have grown fond of this Marshall now, after hating it at first because of it’s really ear damaging treble sound when I first plugged into the Hi channel. Now I realise that the Lo channel suits my guitar, and amps have a Tone control for a reason! It still doesn’t need to be much above 2-3 on the Lo to get a nice tone, else it does start getting too toppy still.
Only on jamming to Robben Ford’s website one night with all the amps set up next to each other so able to compare, I realised this Marshall definitely has its place for Classic blues tone especially on the lead pickup.
As I did so much learning of electronics with this amp also, I had to get it sorted, and I would not have wanted to sell it on with this feature not working properly, as it takes about 45 secs on full Intensity and Speed before the effect would start to occur, and then at a VERY slow rate. Not usable. Here is the Trem section:
As I had changed all the capacitors in this section for new, with the correct values shown, and added the 220pF that was not present in this amp but on the schematic, I thought I would get a comparison from other transistor based Tremolo circuits to try to see a cause of the problem.
I found this one:
Comparing them, which are very similar, with just half the voltage (15V) across this one than the Marshall’s 30V, I thought the place to start would be to remove C23 – the 220pF cap I had added, as this circuit doesn’t have a feedback capacitor from the Collector to Base.
I now fail to see the function of C23 in the Marshall schematic, unless it just alters phase, which is the general principle of the other 3 caps – C20-C22 – each one shifting the feedback waveform phase by 90 or so degrees usually, depending on their value, from what I remember researching positive feedback oscillators at the time of the first repair.
This made no difference at all to the functioning – it was as before.
After some thought, and looking at the really small value Speed control of 5k Ohms in this circuit compared to the Marshall’s 100k Ohm, I thought I would double check both the Speed and Intensity pot values to see if they had drifted markedly over the decades. There were OK.
Changing the 100k Intensity pot was not an option anyway as it forms an integral part of the main voltage divider chain across the whole circuit and is the DC bias point for the input to the gate of the power valve.
This would not affect the function of the Tremolo, as the Tremolo is just slow AC superimposed onto this DC bias anyway. The pot just controls how much AC is superimposed at the R12 gate.
If I had thought further, I would have realised (I did later!) that it makes no difference in the Marshall what value the Speed pot (VR3) was either, as at maximum speed it is at earth, and the oscillation is as fast as it is going to get decided by the 270k R19 anyway, which is still too much resistance as it makes the CR circuit time constant too slow – remember T = R × C from the other Posts?
The larger either component is, the longer the discharge period, the longer the cycle in seconds.
As this new circuit has a only a 220 Ohm resistor with the 5k pot and I need to make the time constant of the Marshall oscillate faster, I thought I would change the seemingly massive 270k Ohm R19 for something less – like half as much at least for a start – say, 100k Ohm and try it.
As this R19 resistor had also been changed for new by me months back (probably unnecessarily looking back with more knowledge), it had long legs on it from a a soldering struggle, so I thought I may as well make life easier and add the 100K in parallel with it by soldering directly to it, which meant not having to remove the board completely either, or replace the 270K completely, and I don’t have to risk damaging the already fragile tracks on this old circuit board. This also meant changing other resistor values would be easier depending on the sonic results of this one. Purist amp techs will be sickened by this I know…
A quick calculation for resistors in parallel gives a value of 1/270k + 1/100k = 1/R or about 73k, instead of the 270k. This means the capacitors can discharge more quickly, so oscillate faster, which is what I need.
Bingo! First time, just about right! The controls were immediately effective, and both fastest and slowest Speed times were really good, and the Intensity actually responsive immediately.
I don’t understand how this value can be so far out from the original design to give the desired results – even other component drift would need to be pretty extreme I would think, for some of the components to have drifted that badly to make this situation so bad in the first place?
Anyway – not complaining – it is a completely useable amp now, and I would be happy to sell it on as such now – though I’m not in a hurry just yet.

Heres the Trem on Daves 1972 Marshall Rev Trem 25W amp: