Hello guys,
Did you use jfets on front end or MPSA18?
If you want the best from symasym, forget MPSA
Did you use jfets on front end or MPSA18?
If you want the best from symasym, forget MPSA
Roender:
Is it okay to use 2sk117 for the inputs,as they have a larger voltage range,if I use 2x43v dc for Symasym?
If not ,any other?
Is it okay to use 2sk117 for the inputs,as they have a larger voltage range,if I use 2x43v dc for Symasym?
If not ,any other?
Yes Ryssen. Try to match them for Idss and if possible at symasym frontend bias current. Don't use degeneration resistors with them
Hi roender,
did you try to bias them a bit higher than 1.5mA? The original bias seems a bit low to me for the fets. You would increase the voltage drop through the bleeder series resistors (22 or 47 Ohms, depending on sym.-version) for a increased current so I did not try to increase the diff-pairs current source yet(using 170'ies as well).
Rüdiger
did you try to bias them a bit higher than 1.5mA? The original bias seems a bit low to me for the fets. You would increase the voltage drop through the bleeder series resistors (22 or 47 Ohms, depending on sym.-version) for a increased current so I did not try to increase the diff-pairs current source yet(using 170'ies as well).
Rüdiger
Hi Onvinyl,
I had a version of symasym with front end cascoded and biased at 2mA on each jfet.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=97144&highlight=
I dropped that design and developed another one with more current on front end, 7ma on each fet, and with folded cascode "VAS". The schematic will be made public as soon I'll get the professional PCB version ready
http://www.diyaudio.com/forums/showthread.php?postid=1313489#post1313489
I had a version of symasym with front end cascoded and biased at 2mA on each jfet.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=97144&highlight=
I dropped that design and developed another one with more current on front end, 7ma on each fet, and with folded cascode "VAS". The schematic will be made public as soon I'll get the professional PCB version ready
http://www.diyaudio.com/forums/showthread.php?postid=1313489#post1313489
Hi,
the 220r across the Vbe sets the CCS current to about 2.9mA i.e.1.45mA through each half of the LTP.
I selected my collector resistors (680r) to better than 0.1% and the collector voltagedrops ended up at 994mV and 992mV, indicating LTP balance <=0.3%. Anatech's matching method works.
the 220r across the Vbe sets the CCS current to about 2.9mA i.e.1.45mA through each half of the LTP.
I selected my collector resistors (680r) to better than 0.1% and the collector voltagedrops ended up at 994mV and 992mV, indicating LTP balance <=0.3%. Anatech's matching method works.
It is a good thing for the fets, but I don't know if it's good for an otherwise unchanged symasym.
Maybe Mike can shine some light on it...
Rüdiger
Maybe Mike can shine some light on it...
Rüdiger
Yes,I wish he could shine..Or roender maybe.
I have som 2SK389,I think I´ll use them instead,they can take 50v,they have kind of odd pinnout. I have to make some adapter.
Ryssen said:
http://www.diyaudio.com/forums/showthread.php?postid=1068995#post1068995
Better change current source with BJT+LED one, for BJT use any TO226 you have
Posting this,if anyone can check that I did this right..
.SCH.zip
An externally hosted image should be here but it was not working when we last tested it.
I´ve seen some use a J-fet even for the Led,any advantage of doing this?
Hi Ryssen,
I like led-BJT current sources. They work really well and I typically design my current sources that way. Using a J Fet doesn't seem to make much sense to me. I imagine you could select some with the exact parameters, but that is a pain. You can simply grab a transistor and use it.
Your current source looks fine to me. You are expecting around 12.5 mA, right?
-Chris
I like led-BJT current sources. They work really well and I typically design my current sources that way. Using a J Fet doesn't seem to make much sense to me. I imagine you could select some with the exact parameters, but that is a pain. You can simply grab a transistor and use it.
Your current source looks fine to me. You are expecting around 12.5 mA, right?
-Chris
Hi,
R12 could probably be increased substantially without any loss of sound quality.
A noise filtering cap across the LED will also help.
I guess you will have about 1.7V across each of the collector loads of the input LTP. Is that voltage (and the second stage currents) the same as in the original?
I suspect that R14 can now be increased to 50k or even 100k, allowing a smaller/higher quality DC blocking cap to be used.
Can someone simulate this and tell us if the open lop gain is similar or lower than the original.
If the OL gain is lower, then the Cdom value (C7) may be reduced following build tests.
R12 could probably be increased substantially without any loss of sound quality.
A noise filtering cap across the LED will also help.
I guess you will have about 1.7V across each of the collector loads of the input LTP. Is that voltage (and the second stage currents) the same as in the original?
I suspect that R14 can now be increased to 50k or even 100k, allowing a smaller/higher quality DC blocking cap to be used.
Can someone simulate this and tell us if the open lop gain is similar or lower than the original.
If the OL gain is lower, then the Cdom value (C7) may be reduced following build tests.
Hmm,my mathematics sucks,but I thought 1,5v(Led)-0,65(VBE)/100(R)=
8,5mA,dpending on what Led is used 1,5-2,0V.
This means 4,25mA on each J-fet in 2sk389,is this to high/low?
According to:http://www.lf-pro.net/mbittner/Sym5_Webpage/symasym5_3.html
the "current" schema,I get it to 0,99v over the collector loads,does this mean I have to increas those reistors?How did you calc.that voltage?
Here I have made the "modifications"Lower R12,and higher R14,and lower C1. 1,59hz must be enough -f3db?
New.sch
LTP currents
Hi,
Assume LED1 voltage is 1.6V.
Subtract 0.6V for Q7 Vbe.
Voltage across R11=1V.
I R11=10mA. This current is very dependant on the LED Vf. It must be checked in circuit.
Assume the LTP is balanced, giving 5mA per half.
Vdrop of R5 & R6 =1.7V
I think you are right to increase the input FET currents, but you have not used the correct ratio of resistor values if you want the second stage currents and gain to be the same as in the original.
However, can we assume that the FETs will give the same open loop gain as the BJT front end? I suspect not.
Extra current through the second stage may change the gain of this stage (higher?) and could restore the open loop gain.
It may be that your choice of resistors will give a well performing amp.
Can anyone tell us what direction the first build should go?
Finally,
the symetrical topology adopted in the Symasym automatically corrects the balance in the two LTPs, making this amplifier much more tolerant of component and voltage changes. It may be this inherent balance that results in good sound.
If this is the case, it points the finger at poor balance of the input LTP as being the possible cause of some amps sounding less than good. D.Self does caution us that balance is important
Hi,
Assume LED1 voltage is 1.6V.
Subtract 0.6V for Q7 Vbe.
Voltage across R11=1V.
I R11=10mA. This current is very dependant on the LED Vf. It must be checked in circuit.
Assume the LTP is balanced, giving 5mA per half.
Vdrop of R5 & R6 =1.7V
I think you are right to increase the input FET currents, but you have not used the correct ratio of resistor values if you want the second stage currents and gain to be the same as in the original.
However, can we assume that the FETs will give the same open loop gain as the BJT front end? I suspect not.
Extra current through the second stage may change the gain of this stage (higher?) and could restore the open loop gain.
It may be that your choice of resistors will give a well performing amp.
Can anyone tell us what direction the first build should go?
Finally,
the symetrical topology adopted in the Symasym automatically corrects the balance in the two LTPs, making this amplifier much more tolerant of component and voltage changes. It may be this inherent balance that results in good sound.
If this is the case, it points the finger at poor balance of the input LTP as being the possible cause of some amps sounding less than good. D.Self does caution us that balance is important
Re: LTP currents
In my experience LED-voltage is ~1.87v at ~2ma.
May i suggest to keep the 2 bjt ccs?
In my experience the 2bjt-ccs has a higher psrr. Also it is thermally stable, as the vbe of the "output device" does not change output current. If the sensing device heats up, the current drops and does not increase like in most other ccs circuits.
You can change the output device to a to126 one, and keep the sensing bjt the small to92. Just keep sure that the sensing device is not slower than the output device, or you risk oscillation.
Yes, without correct adjustment of r5/6, the bias in vas will change.
The jfets have substantially lower gain than bjts, the sk170 is one of the highest gain devices.
Because of the curved transfer curve of jfets, the increased ltp-current will increase gain, but as r5/6 are lowered, the gain is likely to keep much lower than with bjts.
The lowered gain here is not a bad thing, it will increase safety margin for stability.
I also think that the very balanced operation of this topology is the main reason for its good sounding. Also, it features a quite high psrr.
Mike
AndrewT said:
In my experience LED-voltage is ~1.87v at ~2ma.
May i suggest to keep the 2 bjt ccs?
In my experience the 2bjt-ccs has a higher psrr. Also it is thermally stable, as the vbe of the "output device" does not change output current. If the sensing device heats up, the current drops and does not increase like in most other ccs circuits.
You can change the output device to a to126 one, and keep the sensing bjt the small to92. Just keep sure that the sensing device is not slower than the output device, or you risk oscillation.
Yes, without correct adjustment of r5/6, the bias in vas will change.
The jfets have substantially lower gain than bjts, the sk170 is one of the highest gain devices.
Because of the curved transfer curve of jfets, the increased ltp-current will increase gain, but as r5/6 are lowered, the gain is likely to keep much lower than with bjts.
The lowered gain here is not a bad thing, it will increase safety margin for stability.
I also think that the very balanced operation of this topology is the main reason for its good sounding. Also, it features a quite high psrr.
Mike