Today I did some thermal testing with a single JFET (2SK170) CCS dialed in at 0.40mA at 25Celcius. When I warmed it up to 40Celcius its output increased to 0.50mA, which is a bit much.
The LM334 with tempco is far more stable. (0.40mA at 25C -> 0.43mA at 40C).
Thermal fluctuation might be less of a problem with single board test rigs. But in a multi board stack this will definitely need to be considered.
The only solution I can think of is to calibrate the CCS’s at 0.40mA at 40Celcius and take distortion for granted at a cold start of the dac, until it’s at working temperature.
What do you guys make of this?
The LM334 with tempco is far more stable. (0.40mA at 25C -> 0.43mA at 40C).
Thermal fluctuation might be less of a problem with single board test rigs. But in a multi board stack this will definitely need to be considered.
The only solution I can think of is to calibrate the CCS’s at 0.40mA at 40Celcius and take distortion for granted at a cold start of the dac, until it’s at working temperature.
What do you guys make of this?
Excellent data and intresting approach. Quite a sidestep from what others have been doing.
What do you mean with "inspired", did you use your own circuitboard?
"equilibrium", I'm afraid I didn't register that comment, my English/Latin must have let me down. 🙂
But a Single JFET is going to sound horrible at start-up..?
well .... I have feeling that terrible is not exactly appropriate 
put small clip on heatsink , if you are so concerned
put small clip on heatsink , if you are so concerned
Ok that might explain my output bias voltage changing on its own
The problem is when the ccs puts vcom voltage above 4v dc
It will start to noticeably distort
The gap between 400 and 500 ua will certainly do that
The problem is when the ccs puts vcom voltage above 4v dc
It will start to noticeably distort
The gap between 400 and 500 ua will certainly do that
nige2000,
Are you still using your modified PCI-e USB?
Are you satisfied that it is better than the standard USBs?
I bet folks here would be interested in what you did with that if you think it has turned out to be worthwhile.
Are you still using your modified PCI-e USB?
Are you satisfied that it is better than the standard USBs?
I bet folks here would be interested in what you did with that if you think it has turned out to be worthwhile.
Yes still using it
Is there anyone here still using pc audio?
I've tried a few alternatives with mediocre results
Modified Pc with mqn still on top
Second choice would be a rpi or bbb with reclocking and really really clean power
Anyway it's very worthwhile as is lots of other stuff that would seem a little extreme to most guys
as far as I can see or hear there's no one thing to fix sq it's always many things with greater or lesser importance
I guess I am, too!
I think rpi the exception and not the rule!
The player is the key. In my unhumble opinion there is not a player in the class of mqn for linux.
I do not think your explorations extreme in the slightest.
Last edited:
JFET CCS temperature compensation
I tested a 2SK170 to find out how much the source resistance would have to change for a difference between a CCS current of 0.40 vs. 0.50 ma at 2.46v to drain. It's HUGE!
I used a 2SK170 with an Idss of 8.27ma at 9 volts.
The source resistance for 0.40ma was 805 ohms.
The source resistance for 0.50ma was 629 ohms.
I assume that this JFET would need more source resistance to bring the current back down as it heats up. Adjusting the current to 0.40ma at 40C before installation would work. The cold startup current would probably be around 0.32ma, but it shouldn't sound too bad and the shunts would quickly heat everything up.
I found some PTC thermisters, but they don't increase resistance anywhere near as much as is needed to offset the change in JFET current.
One solution is to be able to measure the current through the CCS board while in place and then be able to get a screwdriver to the pot. A stack of DACs could be adjusted this way while at operating temperature.
I hope that the LM334 tempco sounds good and that the temperature compensation be further improved. Perhaps a PTC or NTC thermister in one of the resistances? I am burning in a LM334 tempco circuit and plan to listen to it on my DAC within a week.
I tested a 2SK170 to find out how much the source resistance would have to change for a difference between a CCS current of 0.40 vs. 0.50 ma at 2.46v to drain. It's HUGE!
I used a 2SK170 with an Idss of 8.27ma at 9 volts.
The source resistance for 0.40ma was 805 ohms.
The source resistance for 0.50ma was 629 ohms.
I assume that this JFET would need more source resistance to bring the current back down as it heats up. Adjusting the current to 0.40ma at 40C before installation would work. The cold startup current would probably be around 0.32ma, but it shouldn't sound too bad and the shunts would quickly heat everything up.
I found some PTC thermisters, but they don't increase resistance anywhere near as much as is needed to offset the change in JFET current.
One solution is to be able to measure the current through the CCS board while in place and then be able to get a screwdriver to the pot. A stack of DACs could be adjusted this way while at operating temperature.
I hope that the LM334 tempco sounds good and that the temperature compensation be further improved. Perhaps a PTC or NTC thermister in one of the resistances? I am burning in a LM334 tempco circuit and plan to listen to it on my DAC within a week.
No matter what linux I listen to it lacks detail and is smoothed out to an extreme
Oh If we had the right player for Linux. ....
Carlsor,
I admire your efforts.
If you can get that thing to sound good with a thermister involved I will be surprised.
I admire your efforts.
If you can get that thing to sound good with a thermister involved I will be surprised.
this one looks as even better option than LM334
LT3092
http://www.linear.com/docs/27892
everything covered - voltage window , nice tempco , minimal current
from first glance - preset possible with fixed resistor(s)
LT3092
http://www.linear.com/docs/27892
everything covered - voltage window , nice tempco , minimal current
from first glance - preset possible with fixed resistor(s)
Rick,
First, the LM334 has to sound good without a thermistor. It is NOT a solution for JFET CCS.
I hope that the LM334 tempco can be improved by trying different resistor values.
Thermistors are made of metal oxide which is not a good choice for resistors in audio circuits. These are a last resort only for the LM334. The proof is in the listening.
First, the LM334 has to sound good without a thermistor. It is NOT a solution for JFET CCS.
I hope that the LM334 tempco can be improved by trying different resistor values.
Thermistors are made of metal oxide which is not a good choice for resistors in audio circuits. These are a last resort only for the LM334. The proof is in the listening.
I have heard much posivite with MQN player but have yet give it ago due to the user interface. As for PC routes, I have experimented them all including Audiophile Optimiser) with WinServer 2012 core audiophile edition on dual PC setup with JPlay at audio PC on hibernate mode into WaveIO. Still, I thought I heard the Pi gives a more refine SQ when using with a very clean power, USB wifi and with MPD. Hence, so hopefully for a Botic Cape from Russ. For the meantime, I might give MQN a go this weekend. 😉