Hey folks!
Recently bought myself a single ended E80CC/EL84 amp from a fellow enthusiast. First one I've ever had, and I primarily bought it to use with the Beta 8-loaded Dallas horns I'm building (Follow that bit here).
To test the amp I hooked up some of my fathers' cheap logitech satellite speakers. Whatever was there sounded fine, and now knowing it worked I felt confident connecting the Beta 8's just for fun. After a minute or so, I noticed some low-frequency stuff happening and when I looked at the speaker cones they were oscillating something fierce. I cut power and proceeded to sweat bullets. Let it cool down and switched the tubes around, same thing happened. Went ahead and filmed it, oscillations visible only in the last few seconds.
I asked the guy I bought it off of if he'd ever had any issues of the sort, and he told me no. He'd only built it a few months back and had been running it for extended periods regularly through both his BR-mounted Flat 8's and some 3" Tang Band speakers. Upon request he provided the schematics for the amp -- Some sort of DIY kit he had bought off of Ebay. I'll be posting them below.
Only thing about the schematic that worries me is the absence of any kind of resistor on the screen grid, feel like you usually see those on pentode amps. That said, I'm still fairly new to valve amp design so any input from the sages here would be much appreciated.
In the meantime I'm going to borrow some 100x probes from work and feed it with my signal generator to see if I can make any sense of it. Have plenty of old radio drivers that I don't mind bricking for science.
Cheers,
J
Recently bought myself a single ended E80CC/EL84 amp from a fellow enthusiast. First one I've ever had, and I primarily bought it to use with the Beta 8-loaded Dallas horns I'm building (Follow that bit here).
To test the amp I hooked up some of my fathers' cheap logitech satellite speakers. Whatever was there sounded fine, and now knowing it worked I felt confident connecting the Beta 8's just for fun. After a minute or so, I noticed some low-frequency stuff happening and when I looked at the speaker cones they were oscillating something fierce. I cut power and proceeded to sweat bullets. Let it cool down and switched the tubes around, same thing happened. Went ahead and filmed it, oscillations visible only in the last few seconds.
I asked the guy I bought it off of if he'd ever had any issues of the sort, and he told me no. He'd only built it a few months back and had been running it for extended periods regularly through both his BR-mounted Flat 8's and some 3" Tang Band speakers. Upon request he provided the schematics for the amp -- Some sort of DIY kit he had bought off of Ebay. I'll be posting them below.
Only thing about the schematic that worries me is the absence of any kind of resistor on the screen grid, feel like you usually see those on pentode amps. That said, I'm still fairly new to valve amp design so any input from the sages here would be much appreciated.
In the meantime I'm going to borrow some 100x probes from work and feed it with my signal generator to see if I can make any sense of it. Have plenty of old radio drivers that I don't mind bricking for science.
Cheers,
J
You may need another power supply low frequency decoupling RC network between the first and second stages.
Is there a list of the DC voltages in the circuit?
Is there a list of the DC voltages in the circuit?
My observation is that there are no grid stop resistors in the input tubes and then the c2-r6-r8.. shouldn't that be c2-r8-r6?
My first guess would be that the LF oscillation is caused by the lack of power supply decoupling between the first and the second stage.
The lack of grid stopper resistors is also a bit alarming as both E80CC and EL84 are relatively high gm tubes.
The lack of grid stopper resistors is also a bit alarming as both E80CC and EL84 are relatively high gm tubes.
About 60 years ago, I started experimenting with tubes and transistors. I built a number of 2-stage amplifiers, but every 3-stage amplifier I built oscillated. I was yet to learn about decoupling. The power supply drops with any signal, voltage drops at the plate of the input stage, amplified as positive by the second stage to the output stage grid... and so on. R-C of 10K*100 uF is 1 second (1/2 Hz), power supply capacitors and choke will resonate around 4-5 Hz. As suggested above, a third R-C section would help. Reducing C1/C7 and C2/C8 would help reduce subsonic gain.
Three stages seems unnecessary, given that you can drive an EL84 with a wet noodle. 😉
jeff
Does the low frequency oscillations occur without input signal? I also assume it happens in both channels?
R3 150R, R10 100R. That's a lot of GNFB in the second and third stages. So the first gain stage is needed. No HF dominate pole compensation. So I think error by design for LF stability. Certainly the amp should have decoupling between first and second stage. I would guess for LF stability there needs to be some lead compensation. Without simulation I can only guess and would need OPT details such a primary inductance. The arrangement R2, R3, C4, R10 could do with a topology change. If you want me to simulate let me know.
Where would you suggest i place it? I can identify two filters so far (blue/red). Should i put it in the spot encircled in green?
Unfortunately no. I will, however, be measuring a bit tonight. Since it's a full wave diode recti I assume it's about 240vdc.
Yeah, grid stoppers have to go in methinks. And by PS decoupling you mean additional capacitance between V1a and V1b? Remember reading something about motorboating being caused by parasitic oscillations feeding back through V+.
Don't really understand the effect decoupling caps have in that regard, but I'll take your word for it 😁. By reduce you mean lower value, then?
I was considering simulating it, but I figure measuring on the amp itself should give me the most concrete answers. That said, I haven't measured inductance on the primaries, but i would guess they're in the range of 5-6k. DCR is ~370. Secondaries measure 1.1 Ohms (DCR). Could one use a regular component tester with RL-measurement to check primaries? I have one of those kicking around.
And what would be the proposed topology change for R2,R3,C4,R10?
R11 and C5 decouples the driver stage from the output stage, while R9 is the plate resistor for the input stage.
Something like this should provide adequate decoupling between the driver and the input stage. The component values can probably be calculated but I'd say a few kiloohms and a few tens of microfarads.
Ah, of course. Not used to looking at these things upside down. Tack för förtydligandet!
I should be able to make that happen, but every modification i make to the PCB adds a bit to the jank-factor.
Been considering drawing up an RH84 PCB using SMD components where possible, maybe I'd be better off doing that rather than trying to fix a flawed design...
I should be able to make that happen, but every modification i make to the PCB adds a bit to the jank-factor.
Been considering drawing up an RH84 PCB using SMD components where possible, maybe I'd be better off doing that rather than trying to fix a flawed design...
Hard to tell without having the pcb in front of me but with some luck it should be possible to cut a few traces and insert the extra components without making a complete mess out of everything.
If both R6 and R8 are located close to pin 2 of the tube socket I would leave them as they are, but cutting the traces close to pin 2 and pin 7 on the input tubes and insering stopper resistors there would probably be a good idea.
If both R6 and R8 are located close to pin 2 of the tube socket I would leave them as they are, but cutting the traces close to pin 2 and pin 7 on the input tubes and insering stopper resistors there would probably be a good idea.
One do not have to cut traces to add extra power filtering.
Simply remove R9 and make the suggested extra filters on a small board ( or air build) and
connect these to the connections where the R9 was installed.
BTW i doubt that SMB resistors could deal with tube typical voltages.
Simply remove R9 and make the suggested extra filters on a small board ( or air build) and
connect these to the connections where the R9 was installed.
BTW i doubt that SMB resistors could deal with tube typical voltages.
The placement of R9 makes that a bit difficult, but i should be able to make it work. Won't matter too much if i extend the leads a bit for testing purposes.
One would think so! But there are quite a few 1210-package resistors that can handle up to 300V.
Last edited:
As #13 is suggesting split R9 into two say 22K to C1/plate and 10K to C5/R11 and add 47uF 400V to ground to junction of two. Try this first. If that does not work it will need some simulation. Also check your feedback is not positive. However if this is the case its a very loud motorboating with nearly no audio rather than the other way round.
Low frequency oscillation, Right?
First order of business . . .
There needs to be adequate individual B+ filtering between each stage.
I see 2 main low frequency poles:
C2, and R6 & R8: pole frequency.
V2's plate impedance rp, and the primary inductance of the output transformer: pole frequency.
Two more possible low frequency poles are:
R2 & C4; and R7 & C3.
One more low frequency pole frequency is the loudspeaker woofer.
Generally it takes 3 or more low frequency poles inside a negative feedback loop, to cause the low frequency oscillations.
Note: If R6 is intended to be a grid stopper, it is not.
Connect one end of R6 Directly to the socket grid tab; connect the other end of R6 to C2 and R8. Now, R6 is a Real grid stopper.
Grid stoppers always are related to high frequency oscillations.
The only way a missing grid stopper affects low frequency oscillations, is if the high frequency oscillations caused by the lack of a grid stopper . . .
also starts bursted occillations at a low frequency rate.
First order of business . . .
There needs to be adequate individual B+ filtering between each stage.
I see 2 main low frequency poles:
C2, and R6 & R8: pole frequency.
V2's plate impedance rp, and the primary inductance of the output transformer: pole frequency.
Two more possible low frequency poles are:
R2 & C4; and R7 & C3.
One more low frequency pole frequency is the loudspeaker woofer.
Generally it takes 3 or more low frequency poles inside a negative feedback loop, to cause the low frequency oscillations.
Note: If R6 is intended to be a grid stopper, it is not.
Connect one end of R6 Directly to the socket grid tab; connect the other end of R6 to C2 and R8. Now, R6 is a Real grid stopper.
Grid stoppers always are related to high frequency oscillations.
The only way a missing grid stopper affects low frequency oscillations, is if the high frequency oscillations caused by the lack of a grid stopper . . .
also starts bursted occillations at a low frequency rate.
Yes, it happens on both channels -- although I thought I only saw the left channel oscillate the first time. Could very well be I fixated on it and missed the right channel doing the same thing.
Ok, I'll get around to it tomorrow, hopefully!
Very informative, thank you. Does the lack of a screen grid resistor matter? Like previously stated, I thought EL84s operating in pentode generally had those.