In my PP amp design I've hit an issue with the PSRR of the plate/anode stage of the cathodyne.
Hum on the output is caused by the very low psrr of the cathodyne
Thinking that if I've hit this problem, others must have, I did a little research tonight and found this from Broskie.
Which gives a good understanding of the problem, but in my design the different tubes I'm using for gain and cathodyne don't really lend themselves to this solution.
It was then I designed a modified version of this on paper which is the yellow highlighted parts below:
The addition of a voltage divider formed by R30a and R31a will give us half of the HT and half of the HT ripple at the midpoint, which I've then AC-coupled to the grid of the cathodyne.
My interpretation of this is that it may achieve the same effect as Broskie's solution but without re-working the parameters of the preceding stage.
The screen shot cuts off but the B+ voltage for the cathodyne is 300V and the preceding stage is 275v - this is to get the grid voltage correct for the 6N1 cathodyne.
So to my question. Anyone else tried this trick? I haven't seen it anywhere else which means either of three things:
1) I didn't look hard enough and someone else already did this, or
2) I've overlooked something and this idea won't work, or
3) I've invented something new
Realistically 3 is not likely so that leaves 1 and 2.
Anyone done this before?
Hum on the output is caused by the very low psrr of the cathodyne
Thinking that if I've hit this problem, others must have, I did a little research tonight and found this from Broskie.
Which gives a good understanding of the problem, but in my design the different tubes I'm using for gain and cathodyne don't really lend themselves to this solution.
It was then I designed a modified version of this on paper which is the yellow highlighted parts below:
The addition of a voltage divider formed by R30a and R31a will give us half of the HT and half of the HT ripple at the midpoint, which I've then AC-coupled to the grid of the cathodyne.
My interpretation of this is that it may achieve the same effect as Broskie's solution but without re-working the parameters of the preceding stage.
The screen shot cuts off but the B+ voltage for the cathodyne is 300V and the preceding stage is 275v - this is to get the grid voltage correct for the 6N1 cathodyne.
So to my question. Anyone else tried this trick? I haven't seen it anywhere else which means either of three things:
1) I didn't look hard enough and someone else already did this, or
2) I've overlooked something and this idea won't work, or
3) I've invented something new
Realistically 3 is not likely so that leaves 1 and 2.
Anyone done this before?
For a DC coupled voltage amplifier and cathodyne, the usual recommendation is to have a common B+ to avoid low frequency instability. You should be able to achieve that with some minor adjustment of component values. Also have you elevated the heater voltage? About 60V would be good and might help with hum.
Otherwise what Petertub said makes good sense. It's usually possible to achieve very low levels of ripple with R+C if you have some excess voltage to burn, so that low PSRR is not really an issue. The last cathodyne I did, I supplied HV from a regulator that was supplying the screens of the output tubes, there is no audible hum.
Otherwise what Petertub said makes good sense. It's usually possible to achieve very low levels of ripple with R+C if you have some excess voltage to burn, so that low PSRR is not really an issue. The last cathodyne I did, I supplied HV from a regulator that was supplying the screens of the output tubes, there is no audible hum.
I see the R30a R31a divider making the *same* crap as the R24 V5 stage, but at a far higher impedance. So counting on thumbs, it makes things very slightly worse.
With respect to Broski's judo-think, _I_ think the amplifier designer can usually make *clean* DC inside the amp without trickery, just big/cheap caps. The cathodyne it not a heavy load like a power stage.
With 1VAC on the common(!) power rail: Adding C20a "reduces" the crap at the plate insignificantly, from 0.911 to 0.889. However on inspection, the addition of 1Meg||1Meg loading on your 390k V5 plate load is likely to drop signal gain a lot more than the crap reduction (the "reduction" is incidental to a worse drop of gain).
Adding an R-C filter, 10k and 10uFd, to the 300VDC+1VAC supply, loses 27V (10%) of DC but drops the crap from ~~0.9V to ~0.005V @ 300Hz, 0.000,5V@3KHz, 0.012V@120Hz.
And yes, it looks unwise to run the gain-stage direct-coupled to a cathodyne on different supply nodes. I'n not sure it is sure to be trouble, but it looks like two dancers holding hands across two bobbing boats. When two stages are this intimate, let's be intimate all over.
Adding an R-C filter, 10k and 10uFd, to the 300VDC+1VAC supply, loses 27V (10%) of DC but drops the crap from ~~0.9V to ~0.005V @ 300Hz, 0.000,5V@3KHz, 0.012V@120Hz.
And yes, it looks unwise to run the gain-stage direct-coupled to a cathodyne on different supply nodes. I'n not sure it is sure to be trouble, but it looks like two dancers holding hands across two bobbing boats. When two stages are this intimate, let's be intimate all over.
Yup. DC coupling requires all the components to follow the PS voltage, right? Two power supplies won't move in tandem, save for following the supply voltage.
And correct me if I'm wrong, but if you swap 10k and 10uF for 1k and 100uF, the ripple reduction is the same with 1/10th of the voltage loss?
And correct me if I'm wrong, but if you swap 10k and 10uF for 1k and 100uF, the ripple reduction is the same with 1/10th of the voltage loss?
I've built this one before successfully, the difference is that in the last one I did, the B+ was close to 360 which is getting uncomfortable for the EL84 since I'm running ultra-linear, it means that 360V is also on the screens.
This build, I used a different transformer, which gives me a B+ closer to 330. So the voltage dropper for the Cathodyne is a lower resistance, which seems enough to let the hum through on the anode side.
Tonight I'm gonna play with the power supply voltages a little, see if I can improve it. I've got the gainstage running a lower voltage than the cathodyne to give an easier job matching the voltage on the grid of the cathodyne, since I really wanted to DC couple it.
This build, I used a different transformer, which gives me a B+ closer to 330. So the voltage dropper for the Cathodyne is a lower resistance, which seems enough to let the hum through on the anode side.
Tonight I'm gonna play with the power supply voltages a little, see if I can improve it. I've got the gainstage running a lower voltage than the cathodyne to give an easier job matching the voltage on the grid of the cathodyne, since I really wanted to DC couple it.
In guitar EL84 work, we sometimes have to drop the driver supply WAY down, because the EL84 is so sensitive that a driver eating 90% of G2 supply (slight filter) will cause the EL84s to "faint", grid-block, when working into overdrive.
If you were working 2A3 or such, you might have to figure very close. But for EL84, even EL84 in UL mode, I think you have ample room for simple R-C filtering in the B+ to keep the crap out of the driver.
(Besides which, this puzzle has been solved over and over again....)
If you were working 2A3 or such, you might have to figure very close. But for EL84, even EL84 in UL mode, I think you have ample room for simple R-C filtering in the B+ to keep the crap out of the driver.
(Besides which, this puzzle has been solved over and over again....)
2) No it won't, since your divider is loaded by the output impedance of the preceeding stage. And to correct this you would in effect be doing exactly the same thing as Broskie, i.e. re-working the parameters of the preceding stage.
Thanks Merlin. I actually got the idea from your chapter on cathodynes in which you referenced Broskie's solution. This problem has come back because I've only ever built amps with at least 60 volts B+ headroom over the voltage I want the cathodyne to run at, giving enough decoupling to make this not a problem.
This one however has only 30V, but I have a new idea how to solve this which I'll try later today...
This one however has only 30V, but I have a new idea how to solve this which I'll try later today...
Just use a regulated power supply for the input and splitter. Every amp I've built with one has been dead silent.
Confession
So I found out where the hum was coming from ... I was most perplexed since I'd built this exact design before with no problem. This pointed me to build issues, rather than design issues.
Sure enough I found it.... in the minus rail for the bias. A hairline crack in the PCB track! So the 22µF cap on the step-down (1K decoupled) wasn't connected.
A touch with the soldering iron and everything now works how it should.
I'm sharing this as a cautionary tale. I went down all sorts of rabbit holes trying to come up with an ever-increasingly complex solution to a very simple problem.
Have a laugh at my expense.
Thanks for all the replies and suggestions on here.
So I found out where the hum was coming from ... I was most perplexed since I'd built this exact design before with no problem. This pointed me to build issues, rather than design issues.
Sure enough I found it.... in the minus rail for the bias. A hairline crack in the PCB track! So the 22µF cap on the step-down (1K decoupled) wasn't connected.
A touch with the soldering iron and everything now works how it should.
I'm sharing this as a cautionary tale. I went down all sorts of rabbit holes trying to come up with an ever-increasingly complex solution to a very simple problem.
Have a laugh at my expense.
Thanks for all the replies and suggestions on here.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.