That's perfect and quite reassuring , thank you .
I skip these for now and if needed I try normal ceramics with shortened legs before going SMD route - there seem to be quite few decoupling caps as it is .
How do you match input stage if you don't mind me asking , I have DCA75 but is limited to 12V and transistors work above 15V ?
I skip these for now and if needed I try normal ceramics with shortened legs before going SMD route - there seem to be quite few decoupling caps as it is .
How do you match input stage if you don't mind me asking , I have DCA75 but is limited to 12V and transistors work above 15V ?
They run at 15.4v, I batch them out with the dca to get the best match at 12v, the change in hfe is pretty linear at the Ic they are being run at , but you need to look at the curve traces for Hfe vs Ic and Ic vs Vce to weed out any wrong uns. I usually populate 3 on each side of the input stage and then swap out the 4th on each side until I get the lowest offset.
Matching them judiciously at 12v is usually good enough to get within a few 100mV without the action of the servo. Going the extra mile with the 4th transistor on each side can get you under 100mV, which is as good as it gets,
The hfe varies a lot with temperature, so the final matching in circuit is critical
Matching them judiciously at 12v is usually good enough to get within a few 100mV without the action of the servo. Going the extra mile with the 4th transistor on each side can get you under 100mV, which is as good as it gets,
The hfe varies a lot with temperature, so the final matching in circuit is critical
Transistors matching is something putting me off building another one tbh
If I ever get to it I might try to match them within say 5% and make one of the resistors on each side adjustable, say 2k pot paralleled to 33R and try to adjust to give minimum voltage at the output , measure pot resistance and soldering similar value resistor.
Another thing to consider is transistors working temperature and temperature coefficient, maybe is better to measure Hfe after holding transistors body in the fingers for a certain amount of time as it would bring them closer to working conditions.
If I ever get to it I might try to match them within say 5% and make one of the resistors on each side adjustable, say 2k pot paralleled to 33R and try to adjust to give minimum voltage at the output , measure pot resistance and soldering similar value resistor.
Another thing to consider is transistors working temperature and temperature coefficient, maybe is better to measure Hfe after holding transistors body in the fingers for a certain amount of time as it would bring them closer to working conditions.
One board is finished , test ok but there is one or two things which made bit me worried - when input load unplugged servo voltage goes to 16V and there is fair bit of voltage on the output - I guess this happens as input stays "floating" ?
Second one is the amount of voltage appearing on the output at the start up , not as bad on power down but still , this happens with load resistor plugged in . Anyone could comment on the above .
I bought an oscilloscope but is not with me yet so cannot check for oscillations , second board next weekend opefully
Second one is the amount of voltage appearing on the output at the start up , not as bad on power down but still , this happens with load resistor plugged in . Anyone could comment on the above .
I bought an oscilloscope but is not with me yet so cannot check for oscillations , second board next weekend opefully
Thank you ,
I guess best way to get rid off voltage spike at the startup would be to use muting circuit and short output to the ground for several seconds .
I have bit of voltage fluctuation on the output - 40mV or so but this is in an open air, I see what happens when boxed up .Voltage on the servo is quite high , around couple volts , I should try to halve 150k resistors and see if drops , quite unexpected as input stage seems to be nicely matched and all voltages on both sides are almost the same .
I guess best way to get rid off voltage spike at the startup would be to use muting circuit and short output to the ground for several seconds .
I have bit of voltage fluctuation on the output - 40mV or so but this is in an open air, I see what happens when boxed up .Voltage on the servo is quite high , around couple volts , I should try to halve 150k resistors and see if drops , quite unexpected as input stage seems to be nicely matched and all voltages on both sides are almost the same .
Pete, you need to set the input offset trimmer with a loading resistor and a cart plugged in. Set with a 1k load, leave it a few minutes then reset with 100r.
Unplugging the loading resistor with it switched on can cause the rails to crash very briefly, so you can see the full rail voltage on the output, or at least as much as the servo is correcting albeit very briefly.
If you've matched the transistors super tightly, in input stage and the mirrors, this will be reduced significantly. It's possible to match them accurately enough so the servo has to do almost nothing, just correct for a few hundred mV due to temp change.
If the leds on one rail dim slightly when you unplug the cart loading resistor then that's good indication that matching could be better.
And worth noting, it does take the circuit about 70 hours to settle down from first turn on. I know that sounds like snake oil, but it's not. I've built 41 paradises now, it's the same every time, the big lytics on the mirrors take ages to fully form and dc offset measurably decreases in swing as this happens. Mine usually start swinging +/- 300mV or so, and end up sub mV, once they get boxed up.
Unplugging the loading resistor with it switched on can cause the rails to crash very briefly, so you can see the full rail voltage on the output, or at least as much as the servo is correcting albeit very briefly.
If you've matched the transistors super tightly, in input stage and the mirrors, this will be reduced significantly. It's possible to match them accurately enough so the servo has to do almost nothing, just correct for a few hundred mV due to temp change.
If the leds on one rail dim slightly when you unplug the cart loading resistor then that's good indication that matching could be better.
And worth noting, it does take the circuit about 70 hours to settle down from first turn on. I know that sounds like snake oil, but it's not. I've built 41 paradises now, it's the same every time, the big lytics on the mirrors take ages to fully form and dc offset measurably decreases in swing as this happens. Mine usually start swinging +/- 300mV or so, and end up sub mV, once they get boxed up.
Last edited:
I admit I was bit too eager to plug it in and I forgot about input load resistor , that’s how I realised there is lot of voltage on the servo . I started with 10k resistor and when plugged one side of leds dimmed but it was before I trimmed the input to 0, now when unplugging input resistor all leds are lit all the time .,
I have only one channel ready and plugged 6R8 as its close to impedance of Kontrapunt B I’m using atm , output swings in tens of mV but there is couple of V on the servo , something like 1.5V but all this in open air condition, I try to stick both boards into the shoebox or something similar to check what happens
I try to finish second channel and leave it running for some time , oscilloscope should be with me by then.
What is onboard OS?
I have only one channel ready and plugged 6R8 as its close to impedance of Kontrapunt B I’m using atm , output swings in tens of mV but there is couple of V on the servo , something like 1.5V but all this in open air condition, I try to stick both boards into the shoebox or something similar to check what happens
I try to finish second channel and leave it running for some time , oscilloscope should be with me by then.
What is onboard OS?
The 3 x j310 and 2sk170 onboard output stage.
Under a couple of v on the servo is OK.
I usually build a little cardboard house for the input stage to dial it all in. I'm deadly serious about the 70 hours before it settles down, 10mv in open air is about settled.
I've found that the lower the servo action the better the stereo depth is.
Under a couple of v on the servo is OK.
I usually build a little cardboard house for the input stage to dial it all in. I'm deadly serious about the 70 hours before it settles down, 10mv in open air is about settled.
I've found that the lower the servo action the better the stereo depth is.
Could I have a couple please Simon? They would replace the little bits of cardboard I've had in there for years
