Home
This is only a theoretical point to offer a little new information in the "discussion" concerning valve rolling and equivalents.
It isn't a secret here that I regularly advise people against performing any of the 12Ax7 -> 12AT7 -> 12AU7 -> 12AY7 swaps. These valves are not connected in any way, there is no "12A*7 Family", it just does not exist. Each of these valves was designed to fulfil a completely different role and when one is designed into a stage correctly the others are completely unfitted to work in that stage. This never goes down with guitarists who have no engineering background. Everyone knows that these are "equivalents" and that they offer you "lower gain" in a different model in the same envelope. Well - - - they just don't, that's engineering fact.
The argument is not about tone. There is no point in saying that when a swap is performed the sound will not be as good. It may not be as good to me but it may just be better to the owner and that is final, he prefers it. The argument is about how badly the stage is now out of kilter and what that means in terms of reliability. I recently had a lively discussion with some engineers on a different technical site and I offered this following information to the pot. It stopped all argument in its tracks. It isn't anything brilliant in terms of knowledge or my skills, any technically interested person could do this if he wanted to, but it is at least now done and should make the point that there is more to the swap than the non-technical person is qualified to argue about. The swap was from a 12AX7 to a 12AU7 in a Fender style stage and the consequences of that.
There are other very significant changes in the stage. Static DC anode voltage drops from 202V to 87V. This means that, from a position where you had almost 100V signal swing in both directions before hitting the top rail at its top end, you now have only about 80V before the valve cuts off at its lower end. Now these changes may be more to do with "tonal" considerations but they do show how the stage is now completely undefined in terms of the original design, and of course tone. Why did you buy an amp that you want to make such significant tonal changes to in the first place? You should perhaps have just bought a different amp!
Reliability is really a strong question here. The usual reason for changing out valves is that "I want the gain to drop". There are other much better (safer?) "mods" that can be made to get that without changing valves or even upsetting the stage in any way but these are only for those who can solder and work on high voltage equipment safely. They are not difficult to get to grips with if you can do that.
1. In a bypassed cathode as found in higher gain stages, you could remove the bypass capacitor completely. This will give you a stage with no DC parameters changed, the stage has exactly the same DC current and voltages as before. The gain will have now dropped significantly but the lower cut off frequency will have also moved downwards, (widened) too. That is probably not a problem tonally but you may not be happy with the idea, some high gain players value their sparse dry bottom ends.
2. In a bypassed cathode stage, if removing the cap completely is too much reduction in gain then simply add in a resistor to the cathode bypass leg which, because of the bypass capacitor, is only effective at AC. The cathode resistor along with the anode resistor sets the DC conditions for the stage as before so once again nothing changes for DC. The bypass cap makes the effective value of the cathode resistor at AC very low, (effectively 0ohms for an imaginary perfect cap). For AC signals with the new resistor added, it is now no longer just a cap in parallel with the 1.5k Rk, it is a higher resistor value in parallel so moves the total value from 0ohms upwards as you increase the new resistor value. This in turn decreases the AC gain of the stage without affecting DC conditions in any way.
You would need to disconnect and lift one leg of the bypass cap and add in a resistor in series, so connected between the exposed cap leg and its original hole in the PCB. The value needed will depend on the existing resistor values and how much of a drop in gain you want. Higher values of resistor will give less reduction in gain. (Edit: EEEK! I've just spotted a terrible gaff there. That should have been completely opposite. Higher values of resistor give MORE reduction in gain. Sorry for getting that wrong.) You will need to experiment to get a sensible value but the cathode is a low voltage area (in a very high voltage environment!) so components there are generally not too stressed.
3. In a non-bypassed cathode or just as an alternative drop the value of the anode load resistor. Valves do NOT amplify voltages. They produce a varying anode current which is controlled by the voltage on its grid. The voltage on the grid doubles and the anode current doubles, (roughly). It is the job of the anode load resistor to turn that current into an output voltage. Considering the 12AX7 valve with 100k anode resistor in the example above and replacing it with a 50k resistor (of the correct wattage), the stage gain will drop from about 32x to 19x with a single cheap 10p component. Theoretically the drop should be by half but valves are non-linear, (that's why most people love them, that means they distort), so the theoretical value and the real life value will always differ a little. The dissipation in that resistor is about halved too. The dissipation in the valve will rise to match but is still well within the 1W maximum anode dissipation of the 12AX7. You have now changed the output impedance of the stage and lowered it so the stage frequency response should widen a tad at the bottom end but that really isn't so much as to be noticeable.
4. If you were the right type, (engineering in your blood), you could do this and compensate the other components to get the stage back to its original state. A drop of Ra to 50k needs a raise in Rk to 2k to drop the anode current slightly to exactly the same. To keep the bottom end cut-off frequency the same, as the output impedance has lowered from around 40k to around 30k, would require a change of coupling cap to a very slightly lower value. The change is so slight that I wouldn't worry about that last point personally.
It's just a thought but there ARE alternatives to swapping out valves. Sure, it is an easy exercise to simply pull one and push in another but the changes you make are not all to the good and, despite what the online gurus tell you, there are genuine reliability questions to be answered if you do it. There is always more to the engineering of a piece of kit than it is possible to know if you aren't in the field. You might not be one of the unlucky few but, when you needn't use that particular route, you'll feel a bit pished if you are.
Oh, and I had better make clear, this is a theoretical thread to explain to most why this valve rolling is not a good idea for guitar amps. It is not a suggestion that you try these mods in your H&K amp. These are not viable mods in the case of most of the H&K amps for two major reasons. Firstly, simply because of the inaccessibility of the PCBs which are very close fitted inside. Secondly, because the cathode circuits already contain switched additional components and removing bypass caps or adding resistors is not really sensible with so much else fundamental circuitry which will be upset.
It isn't a secret here that I regularly advise people against performing any of the 12Ax7 -> 12AT7 -> 12AU7 -> 12AY7 swaps. These valves are not connected in any way, there is no "12A*7 Family", it just does not exist. Each of these valves was designed to fulfil a completely different role and when one is designed into a stage correctly the others are completely unfitted to work in that stage. This never goes down with guitarists who have no engineering background. Everyone knows that these are "equivalents" and that they offer you "lower gain" in a different model in the same envelope. Well - - - they just don't, that's engineering fact.
The argument is not about tone. There is no point in saying that when a swap is performed the sound will not be as good. It may not be as good to me but it may just be better to the owner and that is final, he prefers it. The argument is about how badly the stage is now out of kilter and what that means in terms of reliability. I recently had a lively discussion with some engineers on a different technical site and I offered this following information to the pot. It stopped all argument in its tracks. It isn't anything brilliant in terms of knowledge or my skills, any technically interested person could do this if he wanted to, but it is at least now done and should make the point that there is more to the swap than the non-technical person is qualified to argue about. The swap was from a 12AX7 to a 12AU7 in a Fender style stage and the consequences of that.
[The proposal:] In a typical fender input stage (1.5k Rk bypassed, 100k Ra), putting a 12AU7 instead of a 12AX7 results in significant level drop, about 10dB. This is bound to result in a quite different voicing, because of the lower output Z, and the subsequent stages working less hard. I'm not saying it's good or bad, but I would think almost anybody could detect that without any test equipment.
[The Answer:] Don't disagree with any of that. What has it done to the bias current? What has that done to the DC operating point? What has that done to dissipation in the 100k Ra? What has it done to the next stage which is on the other triode in the bottle?
I have just used a simple online calculator for speed, here: Valve Stage Calculator
I'm using your Rk = 1.5k , Ra = 100k, Rload 1Meg, with a grid stopper of 68k and a 300V supply. The Norman Koren 12AX7 model shows Ia = 970uA and Ra=202V. The Norman Koren 12AU7 shows Ia = 2.1mA and Ra = 87V. I would say those levels of change within the circuit conditions are pretty significant. Your argument seem to be "but if it sounds good to the user...". That is absolutely correct and I concur, the user has the perfect right to prefer the sound and choose to adopt that change. I have never questioned that, only as to whether it was wise in the area of circuit design and reliability which is never raised in discussion. The case seems to be made that "it fits in the socket, it doesn't blow up when I switch on, it's good to go." And you pointed out that there is no reliability problem in swapping. However consider this...
300V V+ means that Ra is dissipating 94mW for the 12AX7 and 450mW for the 12AU7. I would call that a significant rise, not disastrous, but one which should be considered. Raising the supply to 400V shows 174mW and 812mW. I would call that a very significant rise, perhaps becoming questionable. If we keep on raising the voltage this does still increase but perhaps that would be unfair. Fender for one seem to specify their Ra resistor as 1W and that seems to be pretty common. I call that - "now it's close to interesting". If Fender had specified an 1/8W 125mW resistor in that place you would consider it stupid. The margin between 94mW / 125mW and 812mW / 1W is pretty much the same!
Questions like that are never raised in guitarists discussions as to which sounds best. This argument is not to do with tone it is to do with the wisdom of making a change which seems to work but is one in which the user has absolutely no understanding of what else has changed and to what extent. That is why many amp manufacturers will immediately invalidate their warranty if it becomes known that this swap has been performed. Even Fender have that written into their warranties. It is often interpreted as "as soon as you change the valves the warranty is invalid". I think, only based on common sense, that it really means "perform this type of swap and the warranty is invalid", for very good reasons. I once spoke to the H&K technical design team about this. Their advice is that while it will not damage H&K amps, they were aware of the propensity to fiddle and checked this out at the design stage, they agreed with me that it should not be assumed it is an acceptable universal substitution for all amps as damage could result in some cases. (They also could not imagine why any guitarist would want to do it as it degrades the tone and flexibility of their amps anyway. But then, they are Germans with death Metal in their DNA so wouldn't understand a gain reduction anyway.)
There are other very significant changes in the stage. Static DC anode voltage drops from 202V to 87V. This means that, from a position where you had almost 100V signal swing in both directions before hitting the top rail at its top end, you now have only about 80V before the valve cuts off at its lower end. Now these changes may be more to do with "tonal" considerations but they do show how the stage is now completely undefined in terms of the original design, and of course tone. Why did you buy an amp that you want to make such significant tonal changes to in the first place? You should perhaps have just bought a different amp!
Reliability is really a strong question here. The usual reason for changing out valves is that "I want the gain to drop". There are other much better (safer?) "mods" that can be made to get that without changing valves or even upsetting the stage in any way but these are only for those who can solder and work on high voltage equipment safely. They are not difficult to get to grips with if you can do that.
1. In a bypassed cathode as found in higher gain stages, you could remove the bypass capacitor completely. This will give you a stage with no DC parameters changed, the stage has exactly the same DC current and voltages as before. The gain will have now dropped significantly but the lower cut off frequency will have also moved downwards, (widened) too. That is probably not a problem tonally but you may not be happy with the idea, some high gain players value their sparse dry bottom ends.
2. In a bypassed cathode stage, if removing the cap completely is too much reduction in gain then simply add in a resistor to the cathode bypass leg which, because of the bypass capacitor, is only effective at AC. The cathode resistor along with the anode resistor sets the DC conditions for the stage as before so once again nothing changes for DC. The bypass cap makes the effective value of the cathode resistor at AC very low, (effectively 0ohms for an imaginary perfect cap). For AC signals with the new resistor added, it is now no longer just a cap in parallel with the 1.5k Rk, it is a higher resistor value in parallel so moves the total value from 0ohms upwards as you increase the new resistor value. This in turn decreases the AC gain of the stage without affecting DC conditions in any way.
You would need to disconnect and lift one leg of the bypass cap and add in a resistor in series, so connected between the exposed cap leg and its original hole in the PCB. The value needed will depend on the existing resistor values and how much of a drop in gain you want. Higher values of resistor will give less reduction in gain. (Edit: EEEK! I've just spotted a terrible gaff there. That should have been completely opposite. Higher values of resistor give MORE reduction in gain. Sorry for getting that wrong.) You will need to experiment to get a sensible value but the cathode is a low voltage area (in a very high voltage environment!) so components there are generally not too stressed.
3. In a non-bypassed cathode or just as an alternative drop the value of the anode load resistor. Valves do NOT amplify voltages. They produce a varying anode current which is controlled by the voltage on its grid. The voltage on the grid doubles and the anode current doubles, (roughly). It is the job of the anode load resistor to turn that current into an output voltage. Considering the 12AX7 valve with 100k anode resistor in the example above and replacing it with a 50k resistor (of the correct wattage), the stage gain will drop from about 32x to 19x with a single cheap 10p component. Theoretically the drop should be by half but valves are non-linear, (that's why most people love them, that means they distort), so the theoretical value and the real life value will always differ a little. The dissipation in that resistor is about halved too. The dissipation in the valve will rise to match but is still well within the 1W maximum anode dissipation of the 12AX7. You have now changed the output impedance of the stage and lowered it so the stage frequency response should widen a tad at the bottom end but that really isn't so much as to be noticeable.
4. If you were the right type, (engineering in your blood), you could do this and compensate the other components to get the stage back to its original state. A drop of Ra to 50k needs a raise in Rk to 2k to drop the anode current slightly to exactly the same. To keep the bottom end cut-off frequency the same, as the output impedance has lowered from around 40k to around 30k, would require a change of coupling cap to a very slightly lower value. The change is so slight that I wouldn't worry about that last point personally.
It's just a thought but there ARE alternatives to swapping out valves. Sure, it is an easy exercise to simply pull one and push in another but the changes you make are not all to the good and, despite what the online gurus tell you, there are genuine reliability questions to be answered if you do it. There is always more to the engineering of a piece of kit than it is possible to know if you aren't in the field. You might not be one of the unlucky few but, when you needn't use that particular route, you'll feel a bit pished if you are.
Oh, and I had better make clear, this is a theoretical thread to explain to most why this valve rolling is not a good idea for guitar amps. It is not a suggestion that you try these mods in your H&K amp. These are not viable mods in the case of most of the H&K amps for two major reasons. Firstly, simply because of the inaccessibility of the PCBs which are very close fitted inside. Secondly, because the cathode circuits already contain switched additional components and removing bypass caps or adding resistors is not really sensible with so much else fundamental circuitry which will be upset.
Last edited by bordonbert on Sat Oct 09, 2021 8:41 am; edited 4 times in total
