I building a 6L6 push pull and have done a couple simulations with PSUD.
in the original williamson circuit, a input capacitor of 8uf is used to keep the output stage alive.
i used PSUD`s constant current stepper to "create" a transient situation and found the voltage to sag a 130v after an increase in current draw from 75ma(idle) to 150ma. isn`t this bad?
i did another simulation with solidstate diodes and large chunky caps with an choke input filter and was surprised to not achieve next-to-perfect "transient" response. my best score was about 10v drop, with 4.5v attributed to losses over the chokes. (which isn`t too bad) but! it cost alot. 2 16H chokes and 1500uf worth of caps.
The most i can afford is a 10H choke input with 1650uf of caps spread out on 22ohm droppers. on ss diodes.
now i`m wondering if a regulator would solve all these problems...?
in the original williamson circuit, a input capacitor of 8uf is used to keep the output stage alive.
i used PSUD`s constant current stepper to "create" a transient situation and found the voltage to sag a 130v after an increase in current draw from 75ma(idle) to 150ma. isn`t this bad?
i did another simulation with solidstate diodes and large chunky caps with an choke input filter and was surprised to not achieve next-to-perfect "transient" response. my best score was about 10v drop, with 4.5v attributed to losses over the chokes. (which isn`t too bad) but! it cost alot. 2 16H chokes and 1500uf worth of caps.
The most i can afford is a 10H choke input with 1650uf of caps spread out on 22ohm droppers. on ss diodes.
now i`m wondering if a regulator would solve all these problems...?
It is not applicable for the Williamson as that amplifier gose in class A, for an amplifier in class AB it is probably quite bad and a power supply for a class AB amplifier must be designed in another way then for a push-pull class A amp.
What kind of transient did you use for your test?, what was the rise time? is that transient giving double current something that would be realistic from music content?
Regarding regulators I think it is possible to do but it is not that simple, as you realise the regulator has to take care of the transient and because of this would not be all that different to a power amplifier with very good performance so you end up with building 2 amplifiers and I dont know if that is a good way to go.
You should also maybe consider that there has been many tubeamplifiers built both in class A and AB without regulation, how did these designers make good sounding amplifiers without regulation and with very small elytics? maybe something to think about.
Regards Hans
What problems? Specs? Which brings me to the very apt remark by Hans
Instinctively I would say that regulation always happens to late since it has to wait for something to act on. Once that something happens it is allready too late. That ofcourse does not necessarily mean that it can't be done (regulation is often credited for tighter bass).
I would like to go even further than Hans by saying get small value caps. PIO or polyprop. 8uF or so and ONE choke. The PP nature takes care of the hum.
I would have liked to ramble more....maybe later.
Cheers,
Bas
i`m actually using the acrosound version.
i think 8uf is terribly under...
i`ve re did my simulations and found stacking caps to do the job just as well. it seems to work well with 4 stages of LCRC filtering.
diodes>20uf>16H choke(30R)>30uf>10R>47uf>15R>68uf.
seems to do well for its cost.
the 1st 20uf is probably going to be a polyester cap, 30uf motor run oil cap, 47 and 68 would be elytics.
i`m operating 360v in class AB1. maxium plate current is 100ma. quiecent is about 88ma(acording to the datasheet).
i`m looking for maxium slam... would this work for me?
i think 8uf is terribly under...
i`ve re did my simulations and found stacking caps to do the job just as well. it seems to work well with 4 stages of LCRC filtering.
diodes>20uf>16H choke(30R)>30uf>10R>47uf>15R>68uf.
seems to do well for its cost.
the 1st 20uf is probably going to be a polyester cap, 30uf motor run oil cap, 47 and 68 would be elytics.
i`m operating 360v in class AB1. maxium plate current is 100ma. quiecent is about 88ma(acording to the datasheet).
i`m looking for maxium slam... would this work for me?
Hi,
Can you please explain what transient you are using for your test, it doesn't make any sense that you get such bad result if the condition is realistic.
Now you say that quicient current is 88mA and max current is 100mA what happened to the doubling of current?
Have you not still considered that all classic amplifiers are designed using low values of caps, have you considered at all that your "test" maybe is not realistic?
BTW I get ~5V drop when stepping instantly from 88 mA to 100mA using 8uF + 8uF and 10H, is that significant when the voltage is 400V? to step instantly i.e 0 rise time is however not realistic and in reality the drop will not have any hearable influence.
Regards Hans
Can you please explain what transient you are using for your test, it doesn't make any sense that you get such bad result if the condition is realistic.
Now you say that quicient current is 88mA and max current is 100mA what happened to the doubling of current?
Have you not still considered that all classic amplifiers are designed using low values of caps, have you considered at all that your "test" maybe is not realistic?
BTW I get ~5V drop when stepping instantly from 88 mA to 100mA using 8uF + 8uF and 10H, is that significant when the voltage is 400V? to step instantly i.e 0 rise time is however not realistic and in reality the drop will not have any hearable influence.
Regards Hans
"If it measures good and sounds bad, -- it is bad. If it sounds good and measures bad, -- you've measured the wrong thing."
Daniel R. von Recklinghausen
What are you simulating/measuring here?
I am not saying you should not measure/simulate. Just trying to find out what exactly it is you are trying to do.
In the meantime I suggest you try a choke input followed by 8uF PIO and listen.
Then build your "diodes>20uf>16H choke(30R)>30uf>10R>47uf>15R>68uf" psu and compare.
Also try "diodes>20uf>16H choke(30R)>30uf" and leave out the lytics.
Cheers,
Bas
(If I am not making sense to you I will stop bothering you and let the engineers step in) 🙂
Daniel R. von Recklinghausen
What are you simulating/measuring here?
I am not saying you should not measure/simulate. Just trying to find out what exactly it is you are trying to do.
In the meantime I suggest you try a choke input followed by 8uF PIO and listen.
Then build your "diodes>20uf>16H choke(30R)>30uf>10R>47uf>15R>68uf" psu and compare.
Also try "diodes>20uf>16H choke(30R)>30uf" and leave out the lytics.
Cheers,
Bas
(If I am not making sense to you I will stop bothering you and let the engineers step in) 🙂
haha... okay....
i did the first "transient" test without knowing the current. it was a whack in the dark. forget about all of that.
why leave out the elytics? without the elytics, i get terrible hum. should i leave the nfb to correct this?
i thought i was always better to have a solid ripple free supply...?
whats the opinion on tuned chokes?
i was reading an article which states that you can cut down hum by many times simply by putting a cap across a choke. is this true. i saw the formulas and i intend to impliment it.
is this advisable?
i did the first "transient" test without knowing the current. it was a whack in the dark. forget about all of that.
why leave out the elytics? without the elytics, i get terrible hum. should i leave the nfb to correct this?
i thought i was always better to have a solid ripple free supply...?
whats the opinion on tuned chokes?
i was reading an article which states that you can cut down hum by many times simply by putting a cap across a choke. is this true. i saw the formulas and i intend to impliment it.
is this advisable?
... swings and roundabouts...
There is a difference in power supply design for Class A circuits Class AB and Class B circuits. And all other Classes come to that.
If we assume that your amp operates in Class A then it's current draw is essentially constant and Bas's suggestions make lots of sense and a small current reserve is needed to minimise ripple i.e filtering action only from the power supply. If the ripple is of only one frequency then tuning an LC filter to minimise the ripple can also help.
However, if your PP amp is in Class B then you might have a 10:1 or more difference between minimum and maximum current draw. This needs much bigger caps in the power supply to supply the instantaneous current draw without the B+ voltage dropping. Under these conditions if the B+ voltage drops then the ripple will increase too. So much bigger caps in the psu for Class B. The LC filter is less effective in this case as more ripple harmonics seem to be generated by the psu...
I suspect your 6L6 PP amp works in Class AB and so is between the two states given above i.e. needs more capacitance than the pure Class A case.
Regulating the psu for a Class AB or Class B amp can have a huge difference on it's sound - absolutely enormous! But the regulated psu must be sized for maximum current draw for best sound quality and then you have an amp that is less efficent than a Class A amp...
Of course the above are only engineering truths and so, for audio, can be ignored 😉 🙄 😉
ciao
James
There is a difference in power supply design for Class A circuits Class AB and Class B circuits. And all other Classes come to that.
If we assume that your amp operates in Class A then it's current draw is essentially constant and Bas's suggestions make lots of sense and a small current reserve is needed to minimise ripple i.e filtering action only from the power supply. If the ripple is of only one frequency then tuning an LC filter to minimise the ripple can also help.
However, if your PP amp is in Class B then you might have a 10:1 or more difference between minimum and maximum current draw. This needs much bigger caps in the power supply to supply the instantaneous current draw without the B+ voltage dropping. Under these conditions if the B+ voltage drops then the ripple will increase too. So much bigger caps in the psu for Class B. The LC filter is less effective in this case as more ripple harmonics seem to be generated by the psu...
I suspect your 6L6 PP amp works in Class AB and so is between the two states given above i.e. needs more capacitance than the pure Class A case.
Regulating the psu for a Class AB or Class B amp can have a huge difference on it's sound - absolutely enormous! But the regulated psu must be sized for maximum current draw for best sound quality and then you have an amp that is less efficent than a Class A amp...
Of course the above are only engineering truths and so, for audio, can be ignored 😉 🙄 😉
ciao
James
Even with original 8uF + 8uF you should get hum below -90dB, if its worse then that I would look elsewhere for the origin of the hum.
The power supply itself gives quite a lot of ripple but it is cancelled by the push-pull stage. You need to consider what kind of amp you are designing, the power supply ripple level that is good enough for a push-pull stage would be really bad for a single ended amp but remember you should always design according to the requirements of the complete circuit.
Regards Hans
Yes it is a Class AB1 amp...
as you said. the current varies and can be a pain in the butt to cut down ripple and sagging lines.
are there reference power supplies you guys use?
or are they any write ups on beefing up a powersupply for a Class Ab amp?
Thanks!!!
as you said. the current varies and can be a pain in the butt to cut down ripple and sagging lines.
are there reference power supplies you guys use?
or are they any write ups on beefing up a powersupply for a Class Ab amp?
Thanks!!!
Class AB? Regulate that rascal.
If you're running the output stage in triode, you need to regulate the main B+ for optimum performance, then use normal RC decoupling for the drive stages (which run class A). If you're going to run the outputs as pentodes, a screen regulator is the main thing you need; you can just use a decent sized e-lytic for the plate feed.
If you want to start more simply, use a simple, non-tuned CLC pi filter, with the first cap sized at 40 uF or so, the second cap in the 100+ uF range. Electrolytics work fine in this position. Leave the esoteric stuff for later, after you've got a working good-sounding amp.
If you're running the output stage in triode, you need to regulate the main B+ for optimum performance, then use normal RC decoupling for the drive stages (which run class A). If you're going to run the outputs as pentodes, a screen regulator is the main thing you need; you can just use a decent sized e-lytic for the plate feed.
If you want to start more simply, use a simple, non-tuned CLC pi filter, with the first cap sized at 40 uF or so, the second cap in the 100+ uF range. Electrolytics work fine in this position. Leave the esoteric stuff for later, after you've got a working good-sounding amp.
Alls I know is there is utterly no hum in the quad 6L6 amp I built which uses +480V plate and screen voltage. Embarassingly simple power supply that would make the hard-core power supply junkies here collapse in shock but screw 'em, it sounds good and works even better. Between idle (40mA per tube = 160mA total) and full signal (um...haven't measured max. current) it drops from 480 to 450V. While delivering 80W into 5 ohms (on the 8 ohm tap.. go figure).
Tim
Tim
fierce ****...😀
okay.. i`ll go back to knocking together a simple powersupply.
a simple CLC would do the job.
i`ll probably regulate the screen with a LM317 floated on zeners. any thing better?
okay.. i`ll go back to knocking together a simple powersupply.
a simple CLC would do the job.
i`ll probably regulate the screen with a LM317 floated on zeners. any thing better?
Screen reg
Use one of the 317 circuits where there's a HV bipolar standing it off, otherwise it's a question of when you lose it rather than if. The circuit that Joe Curcio used to generate driver stage B+ in his Glass Audio ST-70 project is dead simple and can easily be scaled with some heat-sinking to the currents you need for a screen reg.
Use one of the 317 circuits where there's a HV bipolar standing it off, otherwise it's a question of when you lose it rather than if. The circuit that Joe Curcio used to generate driver stage B+ in his Glass Audio ST-70 project is dead simple and can easily be scaled with some heat-sinking to the currents you need for a screen reg.
Hi SY,
Yes, regulate the screens for pentodes! I forgot that completely. Last time I tried that with 807s the amp sounded better in pentode mode than in ultralinear...and in some ways better than triode (but in other ways worse...)
James
Yes, regulate the screens for pentodes! I forgot that completely. Last time I tried that with 807s the amp sounded better in pentode mode than in ultralinear...and in some ways better than triode (but in other ways worse...)
James
Hi,
If you mean that series regulators rely on NFB to work than, yes, they do.
Fortunately regulators, series ones, can do alot more than just hold a voltage constant.
Without a proper PSU even the best regulator is worthless so it's best to understand what's needed for a given application and design accordingly.
If you reject regulators than you could just as well throw out the chokes as well...
Now we all know you guys won't do that, would you?
Cheers,😉
If you mean that series regulators rely on NFB to work than, yes, they do.
Fortunately regulators, series ones, can do alot more than just hold a voltage constant.
Without a proper PSU even the best regulator is worthless so it's best to understand what's needed for a given application and design accordingly.
If you reject regulators than you could just as well throw out the chokes as well...
Now we all know you guys won't do that, would you?
Cheers,😉
i couldn`t find the particular circuit you were refering to but would this fellow work?
An externally hosted image should be here but it was not working when we last tested it.
Better to use a darlington HV BJT than a MOSFET, as the constant forward base voltage (1.2V or so) holds it square over the zener diode, as opposed to the pretty solid but still slightly squishy MOSFET transconductance characteristic. Or add an error amplifier to the MOSFET. Or do the exact same with tubes. 😀
(Something like attached)
Tim
(Something like attached)
Tim
Attachments
- Status
- Not open for further replies.