The attached .PDF file is an amp I'm trying to build. LT-Spice tells me it should at least be functional.
However, during set up the following ocurrs:
1. Rails reach anticipated voltave OK
2. ~20mVdc at output, no AC detectable. OK
3. Nothing gets hot.
Alll would seem to be in order. OK
4. Attempt to set quiescent bias R6 is really a trimpot, but voltage across Q19 sits at exactly 0V, regardless of trim pot position. NOT OK
5. Checked trimpot and Q19. Both test good and PCB traces confirmed correct. Voltage vs. ground collector and emitter .2vdc NOT OK
6. Working backward to Q15 and Q18, I discover that the there is no voltage drop across base and emitter, i.e., they simply don't turn on. Actual voltage vs ground is rail voltage. NOT OK
7. Checked LTPs and tail current -- all is as expected and confirmed by matching against LT-Spice as well
8. Checked all resitor values, diode orientations, transistor pin orientations. All in order.
9. Traced pin-to-pin continuity. No problems.
10. Inspected fot solder bridges, shorts. Found nothing.
11. All transistors check "good".
OK - so what have I missed? What conditions could result in the inability to bias Q15 and Q18? Have I done something fundamentally dumb?
Any suggestions are appreciated.
However, during set up the following ocurrs:
1. Rails reach anticipated voltave OK
2. ~20mVdc at output, no AC detectable. OK
3. Nothing gets hot.
Alll would seem to be in order. OK
4. Attempt to set quiescent bias R6 is really a trimpot, but voltage across Q19 sits at exactly 0V, regardless of trim pot position. NOT OK
5. Checked trimpot and Q19. Both test good and PCB traces confirmed correct. Voltage vs. ground collector and emitter .2vdc NOT OK
6. Working backward to Q15 and Q18, I discover that the there is no voltage drop across base and emitter, i.e., they simply don't turn on. Actual voltage vs ground is rail voltage. NOT OK
7. Checked LTPs and tail current -- all is as expected and confirmed by matching against LT-Spice as well
8. Checked all resitor values, diode orientations, transistor pin orientations. All in order.
9. Traced pin-to-pin continuity. No problems.
10. Inspected fot solder bridges, shorts. Found nothing.
11. All transistors check "good".
OK - so what have I missed? What conditions could result in the inability to bias Q15 and Q18? Have I done something fundamentally dumb?
Any suggestions are appreciated.
sam9 said:
yes you have (sorry couldn't resist) but the problem isn't all that obvious
http://www.diyaudio.com/forums/showthread.php?s=&threadid=16796&perpage=10&highlight=&pagenumber=1
discusses the issue, basically there is no control over VAS bias current with complementary diff pairs + current mirrors - use resistor diff pair load or add some sort of VAS current sense/feedback loop
I recall that thread now, but A-didn't associate it with my current (pun not intended) problem, and B- at the time I had not tried a symetric design so the issue didn't really "stick".
I note the question in the thread "has anyone actually tried to build the Slone design". The schematic above isn't exactly that, but close enough -- and the results seem to answer the question.
I'll see if I can easily mod the PCB to drop the current mirror as I prefer a solution that doesn't add components. That looks like the most promising direction for now.
I note the question in the thread "has anyone actually tried to build the Slone design". The schematic above isn't exactly that, but close enough -- and the results seem to answer the question.
I'll see if I can easily mod the PCB to drop the current mirror as I prefer a solution that doesn't add components. That looks like the most promising direction for now.
Hi Sam,
What is the motivation or advantages for using 2-VAS stages?
I have done some "experimental" models using a complementary VAS and I find that it can be challenging get a stable bias without crossover or heat
and uses more parts. Are you on a mission to seek less idle noise by biasing the VAS in class AB?
What is the motivation or advantages for using 2-VAS stages?
I have done some "experimental" models using a complementary VAS and I find that it can be challenging get a stable bias without crossover or heat
and uses more parts. Are you on a mission to seek less idle noise by biasing the VAS in class AB?Sam,
From the base of Q14 to the supply rail you need at least Vbe(Q14) + Vbe(Q15) = 0,6 + 0,6 = 1,2 V. The current through Q7 – Q10 is about 2mA each. Q5 is turned on so Vce(Q5) is about 0,2 V . This means that the voltage from the collector of Q5 to the supply rail is 0,002 x R10 + Vce(Q5) = 0,136 + 0,2 = 0,336 V. Because the collector of Q5 is connected to the base of Q14, the voltage is even to low to turn on Q14 and moreover Q15.
There are several ways to resolve this problem. The easiest but not the best way is this: connect the base of Q5/Q6 to the collector of Q5 and not Q6. Remove Q14 and R22 and short-circuit the base to the emitter of the removed Q14. The current through Q15 will be about 4 mA. Do also so on the opposite negative side.
But there is more wrong with this circuit. If you look at the input stage (Q7 – Q10) you think this is a low or non feedback amplifier. Then you should expect a resistor from the collectors of Q15/Q18 to ground. But otherwise the configuration is more like a high feedback amplifier.
And what is the function of Q13/R21 and Q16/R25? Protection? I think they will introduce also distortion. Just remove them.
Just let us know what kind of amplifier you want to build.
Before you do the actions above, beware and turn R6 to the highest value.
Marc.
From the base of Q14 to the supply rail you need at least Vbe(Q14) + Vbe(Q15) = 0,6 + 0,6 = 1,2 V. The current through Q7 – Q10 is about 2mA each. Q5 is turned on so Vce(Q5) is about 0,2 V . This means that the voltage from the collector of Q5 to the supply rail is 0,002 x R10 + Vce(Q5) = 0,136 + 0,2 = 0,336 V. Because the collector of Q5 is connected to the base of Q14, the voltage is even to low to turn on Q14 and moreover Q15.
There are several ways to resolve this problem. The easiest but not the best way is this: connect the base of Q5/Q6 to the collector of Q5 and not Q6. Remove Q14 and R22 and short-circuit the base to the emitter of the removed Q14. The current through Q15 will be about 4 mA. Do also so on the opposite negative side.
But there is more wrong with this circuit. If you look at the input stage (Q7 – Q10) you think this is a low or non feedback amplifier. Then you should expect a resistor from the collectors of Q15/Q18 to ground. But otherwise the configuration is more like a high feedback amplifier.
And what is the function of Q13/R21 and Q16/R25? Protection? I think they will introduce also distortion. Just remove them.
Just let us know what kind of amplifier you want to build.
Before you do the actions above, beware and turn R6 to the highest value.
Marc.
Hi !
As jcx already said, you ran into some trap... A trap thanh also stepped into.
Symetrical currentmirror works fine in sims, but not in real world !
At no point the real bias of vas is defined.
You can replace the currentmirrors by simple resistors, or make
the amp asymetrical.
sorry, this amp will be very difficult to get really working,
Mike
As jcx already said, you ran into some trap... A trap thanh also stepped into.
Symetrical currentmirror works fine in sims, but not in real world !
At no point the real bias of vas is defined.
You can replace the currentmirrors by simple resistors, or make
the amp asymetrical.
sorry, this amp will be very difficult to get really working,
Mike
Escape from the trap!
JCX and Mike B: Thanks for the assist. I removed the 4 transistors that made up the two current mirrors as well as R11 and R17, inseting jumpers in theor place. Then I replaced R4, R7 R10 and R16 with more or less appropriate vales (choice limited to what was on hand). First thing was to confirm that the bias could be adjusted -- and it passed that with not problems. I got a clean sine on the 'scope with a dummy load. A preliminary look at THD gets better than the LT-Spice prediction. Next step is confirm everthing by constructing a second copy then do a nmore thorough check out. Bonus is elimination of 6 components!
-----------------------------
Cunningham: My prime motivation is curriosity and self-education. I've built single LTP/ single VAS amps and wanted to play with something different. The IS and VAS were pretty much arrived at by doubling and flipping over the IS $ VAS from something I had built breviously. These particular PCBs will never (probably) end up being built into a nice enclosure but will stay in the garage where I can experiment with things like the Miller compensation, various drive transistors, etc. From the above, you will notice I've already learned one thing -- just regret I wasn't bright enough to figure it out myself.
-------------------------
Marc Vi: Your post gives me quite a bit to think about. You will, above, how I have resolved this initially. When I'm sure I have a stable, functioning circuit as a starting point, I will explore your suggestions.
Q13 & Q16 are, indeed, for protection. f the VI limiter activates, Q15 & Q18 can get get hit with enough current to destroy them. The idea is that before this happens Q13 & Q16 will turn on and steal current from the base of Q14 & Q17, thus turning every thing off. I've measured the potential across R21 and R25 at normal operating conditions and it remains at the uV (or nA) level ensuring that Q16 and Q13 are completely inactive under non-fault conditions. There remmains some testing to be done to be sure they really swith on when needed.
JCX and Mike B: Thanks for the assist. I removed the 4 transistors that made up the two current mirrors as well as R11 and R17, inseting jumpers in theor place. Then I replaced R4, R7 R10 and R16 with more or less appropriate vales (choice limited to what was on hand). First thing was to confirm that the bias could be adjusted -- and it passed that with not problems. I got a clean sine on the 'scope with a dummy load. A preliminary look at THD gets better than the LT-Spice prediction. Next step is confirm everthing by constructing a second copy then do a nmore thorough check out. Bonus is elimination of 6 components!
-----------------------------
Cunningham: My prime motivation is curriosity and self-education. I've built single LTP/ single VAS amps and wanted to play with something different. The IS and VAS were pretty much arrived at by doubling and flipping over the IS $ VAS from something I had built breviously. These particular PCBs will never (probably) end up being built into a nice enclosure but will stay in the garage where I can experiment with things like the Miller compensation, various drive transistors, etc. From the above, you will notice I've already learned one thing -- just regret I wasn't bright enough to figure it out myself.
-------------------------
Marc Vi: Your post gives me quite a bit to think about. You will, above, how I have resolved this initially. When I'm sure I have a stable, functioning circuit as a starting point, I will explore your suggestions.
Q13 & Q16 are, indeed, for protection. f the VI limiter activates, Q15 & Q18 can get get hit with enough current to destroy them. The idea is that before this happens Q13 & Q16 will turn on and steal current from the base of Q14 & Q17, thus turning every thing off. I've measured the potential across R21 and R25 at normal operating conditions and it remains at the uV (or nA) level ensuring that Q16 and Q13 are completely inactive under non-fault conditions. There remmains some testing to be done to be sure they really swith on when needed.
Since this is experimental, redesigns are inevitable. How about removing Q14 & Q17. The darlington-like configuration might make it more difficult to maintain a quiesent current in Q15/18 (VAS) that is just above cut-off but not to much more because the slope of your DC load line for the VAS transistor's is very steep, almost vertical. Perhaps a lower gain (one transistor) could make it more stable. Using less transistors in the signal path is advantagous anyway. Temperature coefficient is going to be a bias factor so don't let the VAS's get too hot. Maybe adding another EF stage so as to raise the impeadence that the VAS must drive, reducing the current and heat. This allows you to put larger emitter resistors on the VAS transistors and lessen the slope of the DC load line. This might also get rid of the need for Q13&16.
You might try to use a multiple turn pot, like 25 turn, to adjust the bias of the differential like a sensetive DC offset control on a single LTP and VAS. By adjusting both at the same time with one pot, might be able to slightly adjust DC bias in VAS.
BTW, current sources are a great and stable way to bias the output EF stage.
Just a few thoughts....
You might try to use a multiple turn pot, like 25 turn, to adjust the bias of the differential like a sensetive DC offset control on a single LTP and VAS. By adjusting both at the same time with one pot, might be able to slightly adjust DC bias in VAS.
BTW, current sources are a great and stable way to bias the output EF stage.
Just a few thoughts....
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.