Some additional Info :
The CRD E452 from Semitec has a nominal current of 4.5mA, but is specified at 3.9mA~5.1mA.
The current mirror formed by Q5,6 has a current gain of 2.44x.
That means the bias of R5b,6b is 9.5mA minimum, and 12.4mA max.
This current feeds both Q1,2 and Q3,4.
For the circuit to work well, the bias current through Q3,4 needs to be at least 1.5mA, ideally > 3mA.
That means the values in the BoM will be fine with Q1,2 Idss of 6~8mA, as those available in the DIYA store.
If you want to use high Idss JFETs (8~11mA from DIYA store), the simplest change to make is to change the CRD to E562.
IMHO high Idss in this position has no real performance advantage except for increased dissipation.
It is also wise to measure the CRD current with 12V supply before soldering, just to check what bias current you are getting.
Patrick
The CRD E452 from Semitec has a nominal current of 4.5mA, but is specified at 3.9mA~5.1mA.
The current mirror formed by Q5,6 has a current gain of 2.44x.
That means the bias of R5b,6b is 9.5mA minimum, and 12.4mA max.
This current feeds both Q1,2 and Q3,4.
For the circuit to work well, the bias current through Q3,4 needs to be at least 1.5mA, ideally > 3mA.
That means the values in the BoM will be fine with Q1,2 Idss of 6~8mA, as those available in the DIYA store.
If you want to use high Idss JFETs (8~11mA from DIYA store), the simplest change to make is to change the CRD to E562.
IMHO high Idss in this position has no real performance advantage except for increased dissipation.
It is also wise to measure the CRD current with 12V supply before soldering, just to check what bias current you are getting.
Patrick
Last edited:
Yes, I would suggest using the JFET Idss range Patrick suggests for this build. If you happen to have matched JFET’s in your drawer, out of range, be prepared for some resistor tweaking. I was able to increase the current through the current mirror to 13.8mA with parts on hand. This enabled Q3,Q4 to work correctly with enough current. The higher current of course made DC offset go out of wack. Installing a 71K5 resistor at R55 dropped offset to -3mV.
As blue pcb's are not my favorite I made my own (and black) boards. Also changed the CM-BJT to HN4A51/HN4C51 which are leftovers from the CM-build. The rest are footprint changes.
Initally I struggled a bit to get the CRD-tail working with the toshibas. I've ended up with 470|1.2K - quite a difference to the BC5xx. Results looks quite ok now.
Q3/4 to GND ~7,2V
Q3/4 ~3,8mA
R58 4,36mA
"Gain" 1 (2.5k hot in - 2.5k Riv), Waveform looks clean
DC offset wandering ~ +/-1mV
@Patrick - thanks for another great diy circuit.
p.s. I've masked the parts/values that haven't been spoilered by Patrick
Initally I struggled a bit to get the CRD-tail working with the toshibas. I've ended up with 470|1.2K - quite a difference to the BC5xx. Results looks quite ok now.
Q3/4 to GND ~7,2V
Q3/4 ~3,8mA
R58 4,36mA
"Gain" 1 (2.5k hot in - 2.5k Riv), Waveform looks clean
DC offset wandering ~ +/-1mV
@Patrick - thanks for another great diy circuit.
p.s. I've masked the parts/values that haven't been spoilered by Patrick
Attachments
Anyone who would take some trouble would be able to work out the BoM and other details from your message and photos above.
That in itself is not a problem.
The problem is when people start to make copies, not all of them fully understand the considerations behind the design choices made,
And then they might end up with modified or "improved" copies which are not necessarily an improvement.
That is what I object to. Especially when some then have the idea to start circulating these copied "designs".
It happened before, even though not in this particular case; I am sure it will happen again.
Patrick
Yes, resitors are 1206 thin film - but not the collector resistors of the bjt's. Didn't have the values at hand so I had to switch temporary to thick films. Thins are 250mW, Thicks are 660mW.
Quick Test with board not fully warmed up:
1,2k 22,9mW (5,2V, 4,4mA)
470 57mW (5,2V, 11mA)
Q5a 2,9mW (0,65V, 4,4mA)
Q5b 27mW (2,45V, 11mA)
~30mW per BJT (300mW max.) - should be ok.
Can't measure temp of the bjt's but way under 55° - maybe 30° Case at ~18° Room.
Quick Test with board not fully warmed up:
1,2k 22,9mW (5,2V, 4,4mA)
470 57mW (5,2V, 11mA)
Q5a 2,9mW (0,65V, 4,4mA)
Q5b 27mW (2,45V, 11mA)
~30mW per BJT (300mW max.) - should be ok.
Can't measure temp of the bjt's but way under 55° - maybe 30° Case at ~18° Room.
Each dual BJT will see ~20mA at 7V, i.e. 140mW.
They are rated at 300mW at 150°C absolute maximum.
I never use them more than 50mW.
It is for good reason that TO92 devices are chosen.
Case temperature (which you can measure) is not the same as junction temperature (which matters).
Cheers,
Patrick
They are rated at 300mW at 150°C absolute maximum.
I never use them more than 50mW.
It is for good reason that TO92 devices are chosen.
Case temperature (which you can measure) is not the same as junction temperature (which matters).
Cheers,
Patrick
Got mine built up finally. Using new values supplied for R1 to 4 as my devices measured similarly to Fran's. Both rails on both channels drawing 20mA. One channel is showing -2.5mV output offset after 10 mins. (3.2 cold) so I'm happy with that channel.
The other is showing 0.00 which I don't believe and it doesn't fluctuate so I think I have an issue there.
Edit....
Which 'out' should we be using if not using the filter. ? Just tried 'out2' on the 0.00 channel and I get an expected reading there..
Left channel can measure DC on both sets of outputs. Seem to behave slightly differently but both settle to 2.5mV.
Right channel ..previously suspect ....0.00mV at GND/Out near the Wima, and a seemingly very stable -2.8mV at GND/Out2.
I observe that there isn't a continuity buzz between the two outputs, but there is across Out and the 3rd via near Out2.
The other is showing 0.00 which I don't believe and it doesn't fluctuate so I think I have an issue there.
Edit....
Which 'out' should we be using if not using the filter. ? Just tried 'out2' on the 0.00 channel and I get an expected reading there..
Left channel can measure DC on both sets of outputs. Seem to behave slightly differently but both settle to 2.5mV.
Right channel ..previously suspect ....0.00mV at GND/Out near the Wima, and a seemingly very stable -2.8mV at GND/Out2.
I observe that there isn't a continuity buzz between the two outputs, but there is across Out and the 3rd via near Out2.