Well, I have a feeling that my heatsinks did act as antennas since I produced some nice smoke last night while hooking up the F5 to music. Everything was biased up fine and tweaked again after 2 hours. When I placed it in my audio chain however, R11 and R12 started smoking. They still measure ok but are somewhat burnt and I will have to replace them. I assume this is due to an oscillation? As far as I can tell, R11 and R12 were the only parts smoking. If not an oscillation, what else could have caused this?
I can simply connect the heat sinks to ground via some wires. Tap the bottom of each sink, and connect those wires to the star ground I have on the copper plate under the amp.
I can simply connect the heat sinks to ground via some wires. Tap the bottom of each sink, and connect those wires to the star ground I have on the copper plate under the amp.
I'll check. That could make sense since I've had other issues with that preamp. If that's the case will I need to add a cap on the input?
not necessarily.
My advice for a DC coupled amplifier is always add DC servo and DC detection and output protection triggered by output offset detection and possibly input mute triggered by output offset.
In all cases, fuse the supply rails to prevent prolonged and excessive DC output current.
since you are "new" around here, you may not have noticed that the wrong kind of signal cable could lead this amp into oscillating
You can add 0.5-1uH air inductor (a few turns of 12-14AWG wire) at the F5 output. I tested it. Even with 0.5uF capacitor as a load it does not oscillate.
And such low value inductor will not affect the sound quality.
But oscillating to the point of smoking out some resistors almost instantly? The signal cables were probably not ideal, but I wasn't expecting smoke. Could it be as simple as shortening the run and using shielded cables? I'm kind of scared to start it up with a signal for fear of doing more permanent damage!
Maybe I'll ground the sinks, use some shielded cables, and try again? Although I'm hesitant to use that tube pre. I only measured a few mV DC offset on it's output, but I would hate to mess things up more if it is the culprit.
Maybe I'll ground the sinks, use some shielded cables, and try again? Although I'm hesitant to use that tube pre. I only measured a few mV DC offset on it's output, but I would hate to mess things up more if it is the culprit.
I don't think that ungrounded heat sinks are a cause of oscillation. I tested my F5 channels w/o a chassis - only heat sink, PCB and PSU, no ground. What are your speakers and what power level you are using? You can add 150-200pF capacitor between ground and JFET gates. F5 is easily oscillating with high capacitive load.
DC offset would tend to blow one rail's worth of devices, not both sides... in most instances... and it would have to be a lot of offset... the speakers would also likely be damaged before the devices died...
My advice is that DC servos are problematic and should be used judiciously - avoided if possible, and very carefully implemented if required. I prefer to NOT loop the DC offset servo line back to the same place that "loop feedback" is connected.
Tinitus, you mean speaker cable, or input Interconnect Cable??
Imho, a very good cap at the input of an amp is fairly benign, and you can always put two input jacks, one AC coupled and one DC coupled - shades of Phase Linear!
Unknown sources always go into the AC coupled input... 😀
_-_-bear
My advice is that DC servos are problematic and should be used judiciously - avoided if possible, and very carefully implemented if required. I prefer to NOT loop the DC offset servo line back to the same place that "loop feedback" is connected.
Tinitus, you mean speaker cable, or input Interconnect Cable??
Imho, a very good cap at the input of an amp is fairly benign, and you can always put two input jacks, one AC coupled and one DC coupled - shades of Phase Linear!
Unknown sources always go into the AC coupled input... 😀
_-_-bear
Nelson Pass covered this... adding capacitance to the gates, iirc, is not a good idea because of the limited current available from the jfets to drive the outputs. He suggests small capacitance in the feedback, if the amp needs stabilization... iirc.
---------------------
Layout is important, what is stable for one layout may be very parasitic for another...
Thanks everyone! Bear, I tend to agree with you on the plan of attack. Adding a cap to the input would be the easiest solution for me. And that's a great idea to have AC and DC coupled inputs. Spectral had AC and DC outputs on their old preamps.
Since grounding the heat sinks is so easy, I might do that as well just to be safe.
Not sure what speakers I'll be using in the end. I had hooked this up to some Dynaudios last night just to test. Of course it never got that far, but they are probably far from ideal for this amp. But that shouldn't matter provided I'm not trying to break any volume records.
Btw, would JFET gate to ground be the best place for the input caps? And is 150-200pF a standard value?
Since grounding the heat sinks is so easy, I might do that as well just to be safe.
Not sure what speakers I'll be using in the end. I had hooked this up to some Dynaudios last night just to test. Of course it never got that far, but they are probably far from ideal for this amp. But that shouldn't matter provided I'm not trying to break any volume records.
Btw, would JFET gate to ground be the best place for the input caps? And is 150-200pF a standard value?
Ok, so it would appear JEFT to ground isn't ideal. Sorry, believe it or not I have probably read 90% of this thread since it started. Of course I've forgotten 90% of that, so I'm forced to still ask questions. Searching this thread for answers isn't exactly easy either. It's quite the behemoth!
Going to get some photos so people can see what I'm talking about and dealing with.
Going to get some photos so people can see what I'm talking about and dealing with.
I think the idea is for an input DC blocking cap ie series input cap. A cap to ground at that point would be for RFI limiting, which you might also want.
Yes. to the above post.
----------------------
The cap discussed earlier was for the gates of the output devices, I think that was the thing the poster referred to...?
_-_-
----------------------
The cap discussed earlier was for the gates of the output devices, I think that was the thing the poster referred to...?
_-_-
I don't think it's limiting the current, it's limiting the bandwidth. Adding such capacitor to JFET gates (not MOSFETs) just introduces simplest RC low-past filter at the amplifier input.
We will have for example 200kHz bandwidth instead of 500kHz, but who cares since it is well beyond audio frequencies.
I tested it. No side effects in measurements or sonically. It reduced square signal overshoot and got rid of any oscillation even with high capacitive load. Alternatively you cad add 5-10nF feedback capacitors to 100Ohm feedback resistor pairs.
Last edited:
I think it has been said it could be both
shielded is said to be safer
and traditional speaker cables
maybe best to avoid very special cable designs
maybe it will not always cause problems, but might
or at least I have understood it that way
What value am I looking at for a series input cap then? Will 2.2uF work? I have some here. I tried to calculate how that would roll of the response, but wasn't getting realistic numbers.
My very incomplete understanding is that the gates of the Mosfet are capacitive already, so you have to be able to charge and discharge the capacitance "quickly" so that bandwidth can be preserved. In practice this means that one has to source and sink current to discharge the capacitance and charge it, adding capacitance when the available current in the driver is limited then slows the circuit down, perhaps asymmetrically... so not a great idea in this circuit. The impedance looking back at the driver starts to look important with Mosfets, especially if you are trying to drive a row of them.
NP suggests the feedback caps as you noted...
_-_-bear
Should be fine...
It's actually a network of the output impedance of the preamp/source, the series cap and the resistance to ground at the input of the amp... it is a good idea to
put another high value resistor at the "open" side of the cap so that it will drain off any accumulated DC charge. You don't want to discharge that cap into something that you plug in to the input jack... One can make the case of putting the "input impedance setting" resistor before the cap, so that the preamp sees a resistive load in the main, and the highish value resistor on the other end, near the gate of the jfet... a bit depends on the scheme used in the case of a DC and AC input being implemented...
_-_-bear