ac ripple on screen grids Ultralinear

[Rattler66]

hey guys

i have a ef86 preamp into a 12ax7 pi into a 2xkt66 UL power amp. Currently using a 5u4gb rectifier but will be going back to GZ34. It has a horrible buzz/hum in the power amp. i've narrowed it down to AC ripple on the screens. i isolated the power amp and it still hummed as soon as i ac ground the screens the hum went away. my plan of attack is to add another rc filter before the plate/screen node to clean up the ripple, before it hits the plates.

At max current draw (depending on bias) I’m seeing 430 ma (I put my meter in series between the standby switch and the first filter node and cranked the bias). I plan on running to a 60u filter cap then two 100 ohm 10 watt wirewound resistor in parallel then to my original plate/screen node.
According to my math 430ma X 50 ohms =21.5 volts
Each leg will take 21.5 volts x 215ma for a total of 4.7watts dissipation per resistor (well below its max wattage)

I chose this method because I was able to get the power resistors locally for 2 bucks. But I am open to another method that would be better. Any help would be appreciated. thanks

[loudthud]

Measuring current before the first filter cap won't give a accurate picture of what's going on. If you put a 1 Ohm resistor in series with the transformer center tap, a scope will show you what the peak of the current pulse is.

A CRC filter won't give enough ripple suppression without a huge Voltage drop. A small choke will do a much better job and won't stress the rectifier tube as much, mostly because you don't need large filter caps. Something around 1 Henry (+/- 50%) with a DC resistance of 50 Ohms or less and rated for DC current of at least 300mA will do the job. Hammond 159V, 159Y will work. Triode Electronics C354 is marginal but will probably work. Link: C354 Dynaco Choke Replacement also Sunn 3-10C MADE IN USA Sunn 60W amps used GZ34 - 20uF - C354 - 30uF.

[pdf64]

I suspect your diagnosis (screen grid ripple).
Adjusting the bias so as to allow a massive HT current draw seems a bad idea period, and does not replicate large signal conditions.
What is the HT ripple at a reasonable idle state?
Does it then hum if operated with the phase splitter removed?

[Rattler66]

Measuring current before the first filter cap won't give a accurate picture of what's going on. If you put a 1 Ohm resistor in series with the transformer center tap, a scope will show you what the peak of the current pulse is.

A CRC filter won't give enough ripple suppression without a huge Voltage drop. A small choke will do a much better job and won't stress the rectifier tube as much, mostly because you don't need large filter caps. Something around 1 Henry (+/- 50%) with a DC resistance of 50 Ohms or less and rated for DC current of at least 300mA will do the job. Hammond 159V, 159Y will work. Triode Electronics C354 is marginal but will probably work. Link: C354 Dynaco Choke Replacement also Sunn 3-10C MADE IN USA Sunn 60W amps used GZ34 - 20uF - C354 - 30uF.

I think you're absolutely right as of now this is my next line of attack. Standby I will give some more info in my next post

[Rattler66]

I suspect your diagnosis (screen grid ripple).
Adjusting the bias so as to allow a massive HT current draw seems a bad idea period, and does not replicate large signal conditions.
What is the HT ripple at a reasonable idle state?
Does it then hum if operated with the phase splitter removed?

Thanks, yes it hums with the PI removed. I went as far as breaking the connection right after the coupling caps so as to leave the power amp alone and it still hummed. only ac grounding the screens shut it up.

As far as the HT ripple, do I just switch my meter to ac and measure from screen to ground? I made have done it wrong in the past and couldn't get a steady reading.

[Rob's Radio-Active]

Using the meter on its AC setting should give you an good idea of the ripple, though it won't be absolutely accurate due to the meter likely being calibrated for a sine wave and ripple is more of a sawtooth wave...but it should be close. If your meter is a "True RMS" type, then it should be accurate. Be careful of that high voltage on those screens. I usually measure AC riding on DC (using a 10 megohm DMM or VTVM) through a 0.1uF capacitor rated at a much higher voltage than the DC plus AC peak. That way I can be sure the meter isn't fooled by the presence of DC. It is also just a bit safer, in my opinion...though that capacitor has to be discharged before handling it.

cheers
Rob

[Rattler66]

Using the meter on its AC setting should give you an good idea of the ripple, though it won't be absolutely accurate due to the meter likely being calibrated for a sine wave and ripple is more of a sawtooth wave...but it should be close. If your meter is a "True RMS" type, then it should be accurate. Be careful of that high voltage on those screens. I usually measure AC riding on DC (using a 10 megohm DMM or VTVM) through a 0.1uF capacitor rated at a much higher voltage than the DC plus AC peak. That way I can be sure the meter isn't fooled by the presence of DC. It is also just a bit safer, in my opinion...though that capacitor has to be discharged before handling it.

cheers
Rob

Thanks, if I get a chance tonight I will look into it. i'll post back, thanks again

[Rattler66]

I didn't get a chance to read the ac ripple, I may later tonight. Anyway my power transformer is a 375-0-375 50 watt PT, after rectification i'm seeing about 490vdc at the plate/screen node the output tranny is a heyboer jtm 45 radio spares 6.6k UL. is there a possibility that the pt is just too big and causing the excessive ripple. this amp is a dr z route 66 type amp and his amp has the same config yet apparently no hum issues, thanks

[trobbins]

Have you checked the idle bias in the output tubes, and are they reasonably matched?

Is the output stage fixed or cathode biased, and what idle bias power level are you targeting? Fixed bias can be a common hum source.

Measuring the AC ripple voltage on the B+ feeding the OT CT is the preferred measurement point - measuring at an anode or UL screen may give ambiguous readings.

How are you 'measuring' the hum/buzz - using a speaker, or a fixed load resistor with a voltmeter?

As a temporary measure, or even a permanent one if not aesthetically displeasing, I have used recycled flourescent light ballasts for that rating choke (1.8H at >200mA DC for 240VAC 40W ballast).

[potatofarmer]

A choke before the plate supply is the better answer (for fidelity, certainly) and if you check out old Sunn/Dynaco schematics you can see they did just that, sometimes using a parallel pair of chokes for increased current handling.

If you look at the schematic for any of the 135W fender twin / bassman 135 or 70, fender solved this issue by using a bias balance control, which nulls the ripple at the OT... until the amp is overdriven. I've greatly enjoyed the overdriven tone of my 135W'er, but I've never compared it with using a pi filter before the plate supply. Certainly a cheaper solution, at least.

[trobbins]

If you look at the schematic for any of the 135W fender twin / bassman 135 or 70, fender solved this issue by using a bias balance control, which nulls the ripple at the OT... until the amp is overdriven.

That bias balance pot varies the idle bias current on each side of the PP stage, so that the OT can be set up to operate with negligible DC current and hence achieve better low frequency response through the OT. As a by-product, the AC voltage across the OT half-windings should be similar (as the effective resistance from the anodes, and screens, should be similar at idle), and hence provide some cancellation of residual hum that could couple to the speaker. The transformer also has coupling capacitance from primary to secondary, which may also contribute some residual hum if the OT B+ has some ripple. Amps with feedback from the speaker output will normally attenuate those residual hum sources that couple in to the output stage.

[Rattler66]

Have you checked the idle bias in the output tubes, and are they reasonably matched?

Is the output stage fixed or cathode biased, and what idle bias power level are you targeting? Fixed bias can be a common hum source.

Measuring the AC ripple voltage on the B+ feeding the OT CT is the preferred measurement point - measuring at an anode or UL screen may give ambiguous readings.

How are you 'measuring' the hum/buzz - using a speaker, or a fixed load resistor with a voltmeter?

As a temporary measure, or even a permanent one if not aesthetically displeasing, I have used recycled flourescent light ballasts for that rating choke (1.8H at >200mA DC for 240VAC 40W ballast).

The idle bias isn't perfectly matched maybe up to 10ma difference, but I've been through different sets of tubes and have gotten it down to about 3-4ma difference and still humming. I'm biasing 70%-17.5 watts. Not measuring hum just hearing it through the speaker.

[Rattler66]

A choke before the plate supply is the better answer (for fidelity, certainly) and if you check out old Sunn/Dynaco schematics you can see they did just that, sometimes using a parallel pair of chokes for increased current handling.

If you look at the schematic for any of the 135W fender twin / bassman 135 or 70, fender solved this issue by using a bias balance control, which nulls the ripple at the OT... until the amp is overdriven. I've greatly enjoyed the overdriven tone of my 135W'er, but I've never compared it with using a pi filter before the plate supply. Certainly a cheaper solution, at least.

bias balance control was a thought, just wasn't sure if indeed was a possibility. I may possibly try that out also.

[Rattler66]

measured the ac ripple

okay I connected my meter to b+ right at the plate/screen node. the reading fluctuated until I hit the hold button a few times and it would keep settling on 28vac ripple.. does that sound right. my meter says it's TRUE RMS.

now I connected the meter to the screen and I got 1.5vac and as I increased the bias it started to hum and climbed into the 20's vac. so there definitely is ac ripple as I increase the bias which is exactly when it starts to hum and the plate voltages are erratic

[trobbins]

I think that means you have about 450V B+ and 40mA for about 18W plate dissipation per KT66. Using PSUD2 indicates the B+ ripple voltage should be about 12Vpp, or circa 4Vrms, with a 60uF cap and that power transformer secondary and a 5U4G.

Sounds like you have a bad part, or wiring, or your measurement technique is not right.

If your caps can withstand 550VDC, or you have a variac, then you can firstly check the unloaded B+ for DCV and AC ripple, and a steady reading.