OK, at this point I would suggest you to up the bias a bit (yes, up
, to around 312,5mV over emitter resistors, so each output transistor will dissipate 200mW at idle. Now, your headamp will stay in class A most of the time and will last forever. Just replace R19/49 with, say 22R resistor and go higher until you get desired 312,5mV over emitter resistor. 
I think R11 and R15 (plus trimpots) are meant for biasing the output stage, Although i could aways be wrong, when you got me to short the resistor the voltage drop across the output resistors never changed. Third, I don't think it's safe to increase the heat dissipation on the output transistors, they already idle at ~85c, I could add TO-220 heatsinks to the tops of them, two transistors per a heatsink.OK, at this point I would suggest you to up the bias a bit (yes, up
What about overcompensating the the feedback circuit?
Directly no...indirectly yes...circuit DC voltages/currents will be influenced a somewhat by voltage drop over R1/2/8/9 if you change value of R11/15.
That is simply impossible...few post above you sad you have measured ~95mV and I would call it a massive drop from a previous ~960mV.
With +/-16VDC supply, Pd of 200mW for TO-92 and 12,5mA of current they are perfectly safe...read the datasheet.
There are two 1N4841 diodes in series with R19, I should have measured the voltage drop across them, This whole thing has been going down hill and I apologize for my poor performance. I've still been fighting with this amp, and just about on the verge of giving up.
I still have a over biasing problem, the transistors are ~90-100c, I'll handmake some simple heatsinks made out of basic aluminum, using thermal epoxy to bond them to the output transistors.
As for biasing the output stage, I have two options:
1: I can remove D3 and D4 and replace R19 with something like a ~500 ohm resistor.
2: I can use aparatusonits's method and replace R19 with a >22 ohm resistor
(On top of this, I might spend some time to find some 1.6volt LED's, since 1.65v is too much, i think...)
I must also deal with a injection of noise from the mains courtesy of my computer's PSU. I don't know if i should simply add a simple ~0.01uF cap across neutral and hot, or go with a full blown HF filter.
I'm almost considering to just remake the whole board in eagle, but i think that would be a little rash. I just feel as though i regret this.
I still have a over biasing problem, the transistors are ~90-100c, I'll handmake some simple heatsinks made out of basic aluminum, using thermal epoxy to bond them to the output transistors.
As for biasing the output stage, I have two options:
1: I can remove D3 and D4 and replace R19 with something like a ~500 ohm resistor.
2: I can use aparatusonits's method and replace R19 with a >22 ohm resistor
(On top of this, I might spend some time to find some 1.6volt LED's, since 1.65v is too much, i think...)
I must also deal with a injection of noise from the mains courtesy of my computer's PSU. I don't know if i should simply add a simple ~0.01uF cap across neutral and hot, or go with a full blown HF filter.
I'm almost considering to just remake the whole board in eagle, but i think that would be a little rash. I just feel as though i regret this.
Hey...come on...you have some troubles with it, so what...but it plays music so you are 75% done
Do not replace biasing diodes D3/4 + R19 with one single 499 Ohm resistor...Diodes are there to provide some temperature compensation for output stage.
Could you measure DC voltage over R3/4/5/6 and R17/21 for as please?
Forgot to say...while taking measurement short input (signal to signal ground) and without load connected to output.
Do not replace biasing diodes D3/4 + R19 with one single 499 Ohm resistor...Diodes are there to provide some temperature compensation for output stage.
Could you measure DC voltage over R3/4/5/6 and R17/21 for as please?
Forgot to say...while taking measurement short input (signal to signal ground) and without load connected to output.
Last edited:
Sure thing!, I found one of the pots to be broken, these measurements are of the right channel.
R3= 0.14v
R4= 0.11v
R5= 0.17v
R6= 0.12v
R7= 14.54v
R21= -14.56v
I attempted to jumper R19 again, my wire was hard to keep in place, but i managed!
Routput(J)= 0.2v
Routput= 0.912v
22ohm resistor it is!
Out of curiosity (don't spoil this), using your values from above; I "estimated" what would the temperature would be if the amp was down biased to 0.35v.
I estimate 29.4c, I won't need a heatsink if I'm correct
I can't seem to get a recoding of the PSU noise, nor do i know if it's there at the moment, It's clearly above 60hz, but i have no clue how to eliminate it.
Edit, Post #21 is wrong
R3= 0.14v
R4= 0.11v
R5= 0.17v
R6= 0.12v
R7= 14.54v
R21= -14.56v
I attempted to jumper R19 again, my wire was hard to keep in place, but i managed!
Routput(J)= 0.2v
Routput= 0.912v
22ohm resistor it is!
Out of curiosity (don't spoil this), using your values from above; I "estimated" what would the temperature would be if the amp was down biased to 0.35v.
I estimate 29.4c, I won't need a heatsink if I'm correct
I can't seem to get a recoding of the PSU noise, nor do i know if it's there at the moment, It's clearly above 60hz, but i have no clue how to eliminate it.
Edit, Post #21 is wrong
Last edited:
From your measurement I can tell you that you have under-bias input stage and over-bias VAS stage. I think, if your resistors in circuit are all correct in value, that you have problems with R10/14 100k trimers...take them out of the circuit, you don't need them.
Disconnect one end of R61/1Meg resistor, pull servo opamp out of circuit (you should use J-fet opamp for servo opamp e.g. LF411 or similar, not BJT input such as OP27),
pull lower end of R12 and upper end of R13 resistors of the pcb and temporally solder both ends together in "air".
Short the input signal to signal gnd, no load at output and measure voltage again over all resistors.
Disconnect one end of R61/1Meg resistor, pull servo opamp out of circuit (you should use J-fet opamp for servo opamp e.g. LF411 or similar, not BJT input such as OP27),
pull lower end of R12 and upper end of R13 resistors of the pcb and temporally solder both ends together in "air".
Short the input signal to signal gnd, no load at output and measure voltage again over all resistors.
I'll measure all the voltages tomorrow and (completely?) remove the DCservo circuit.
I was using ~600ohm resistors for R11 and R15, I need to use the 5K trim pots that I installed. So here are the measurements of having R11=499ohms and R15=499ohm (pot and resistors).
R3= 0.213v
R4= 0.183v
R5= 0.220v
R6= 0.178v
R17= 14.50v
R21= 14.48v
Here are the measurements of having a zero DCoffset
R3= 0.206v
R4= 0.182v
R5= 0.235v
R6= 0.187v
R17= 14.52v
R21= 14.46v
This is good? I'm using OPA134PA.
I don't know how much more of this i can take, fix some problems, more problems arise! I've never had this much problem with electronics ever.
OK, here we go again!
Decreasing R19 to 24.9ohms makes everything a lot more sensitive, the DC servo works much better now (or it appears to), The amp is a lot more sensitive to the power supply voltages, The LM3xx regulators won't be good enough in this operation.
Adjusting the volume pot will move the offset move a good -100mV (rightchannel) left channel moves less (+/-20mV)
So, I provide myself with a ultimatum
A. I rework the entire PSU to include more regulation, more noise filter ect ect. While also completely redoing the entire boards, buying a bunch of new parts and trying again from from scratch (almost).
B. Giving up on this project, selling all that i have for ~$100CAD and loosing ~$70CAD in the processes and starting something else.
I've been saying this too much, but quite frankly... I don't know what to do...
Cheers
OK, here we go again!
Decreasing R19 to 24.9ohms makes everything a lot more sensitive, the DC servo works much better now (or it appears to), The amp is a lot more sensitive to the power supply voltages, The LM3xx regulators won't be good enough in this operation.
Adjusting the volume pot will move the offset move a good -100mV (rightchannel) left channel moves less (+/-20mV)
So, I provide myself with a ultimatum
A. I rework the entire PSU to include more regulation, more noise filter ect ect. While also completely redoing the entire boards, buying a bunch of new parts and trying again from from scratch (almost).
B. Giving up on this project, selling all that i have for ~$100CAD and loosing ~$70CAD in the processes and starting something else.
I've been saying this too much, but quite frankly... I don't know what to do...
Cheers
How about listen carefully what others and I have suggested so fore and do that exactly, nothing more, nothing else?!
You don't have problems 'cos your PSU is bed or something...you have problems 'cos DC stability of your amp is compromise. As I said before, you don't have enough current in input differentials and you have excessive current in VAS stage. So I ask you again to follow steps given by me few post above. Also, it's normal to see output DC offset drift a little while adjusting volume potentiometer in this design.
Let me check my understanding here.
The current standing is that the inputstage is under biased because of the servo loop. This is my understanding at the moment, Nothing to do with the 1.65volt LED instead of the 1.6volt. I don't even understand the relationship between the voltage drop across the LED and the amount of current flowing through the LED.
Now what I've done is replace a couple resistors, And completely change the circuit. The right channel showed the most peculiar behavior, when both R10 and R14 are exactly 499ohms, the offset is a good +100mV, after i changed the value of 19 to 25ohms. The left channel only changed about 20mV.
This is a significant change, showing the input stage is not correctly biased, I'm all for you on that one. But! I don't understand why the servo loop would cause such a thing, and explanation would help me now and in the future. I want to be able to justify what I'm doing so I don't waste any more parts.
The current standing is that the inputstage is under biased because of the servo loop. This is my understanding at the moment, Nothing to do with the 1.65volt LED instead of the 1.6volt. I don't even understand the relationship between the voltage drop across the LED and the amount of current flowing through the LED.
Now what I've done is replace a couple resistors, And completely change the circuit. The right channel showed the most peculiar behavior, when both R10 and R14 are exactly 499ohms, the offset is a good +100mV, after i changed the value of 19 to 25ohms. The left channel only changed about 20mV.
This is a significant change, showing the input stage is not correctly biased, I'm all for you on that one. But! I don't understand why the servo loop would cause such a thing, and explanation would help me now and in the future. I want to be able to justify what I'm doing so I don't waste any more parts.
No
Voltage drop over LEDs plays significant factor and it's in close relation with the amount of current flowing throught input differentials and everything else.
Will you stop making randomly changes hoping for the best...This is the last time I am telling you to do exactly as you told...or I'll quit for good.
Disable DC servo circuit entirely as explained before, (we will address the servo circuit latter) and take the measurements (voltage drop over all resistors) so we can know what are we dealing with.
I'm done, I'm done with my attitude towards this project, I've had to apologize on several occasions to people because of it. This needs to end, So i self propose to myself to rework the entire project, I need to some kind of work, because melancholy doesn't work.
My plan of action:
1. Design a better PSU pcb and add a couple details. This includes a bigger bank of capacitors, a better CRC network, better regulation and better filtering. Supply tracking is in order. The PSU board will still be hand etched
2. Buying a Pro Dynalo board. Higher quality then the board I have made, Buying this will make sure I don't have to deal with my shotty workmenship.
3. Buying more parts, I'll finally give in and supply the board with how it should be. No pot substitutes (using 100K pots this time), More match resistors (possibly to 0.1%)
I found ebay has many sources for 2SC1815/2SA1015 transistors, namely this source: 100 pcs 2SC1815 2SC1815GR NPN 50V 150ma Toshiba on eBay.ca (item 390112237717 end time 22-Apr-11 21:57:40 EDT)
I can't find if there's been any fakes on the market for these transistors.
I'll buy the proper 1.6volt LED's, I don't know how i ended up with 1.65volt LED's... but i did, and that needs to be fixed.
I need to know if the possibility of fakes is too great.
My plan of action:
1. Design a better PSU pcb and add a couple details. This includes a bigger bank of capacitors, a better CRC network, better regulation and better filtering. Supply tracking is in order. The PSU board will still be hand etched
2. Buying a Pro Dynalo board. Higher quality then the board I have made, Buying this will make sure I don't have to deal with my shotty workmenship.
3. Buying more parts, I'll finally give in and supply the board with how it should be. No pot substitutes (using 100K pots this time), More match resistors (possibly to 0.1%)
I found ebay has many sources for 2SC1815/2SA1015 transistors, namely this source: 100 pcs 2SC1815 2SC1815GR NPN 50V 150ma Toshiba on eBay.ca (item 390112237717 end time 22-Apr-11 21:57:40 EDT)
I can't find if there's been any fakes on the market for these transistors.
I'll buy the proper 1.6volt LED's, I don't know how i ended up with 1.65volt LED's... but i did, and that needs to be fixed.
I need to know if the possibility of fakes is too great.
I don't know what kind of regulated PSU you are using, but for best results you will have to use super-shunt regulator, 'cos your PSU is in series to the load/headphones, so in the end you "listen" your PSU intead amp. Also, your PSU have to be at least fast as the amp or faster (better), low noise and low output impedance.
Please, do that...they are chip and I guess properly design.
Don't vaste you money on 0,1% resistor...1% ones are all you need.
As for original transistors, you can buy directly from Fairchild KSA1015/KSC1815 substitutes (they are good) or try to locate Toshiba genuine ones like here 2SA1015-Y & 2SC1815-Y Audio Amplifier Transistor, x100 | eBay
As for LEDs, just buy enough of them from reliable source and find close Vf (voltage drop at operated current) match ones.
I based the PSU off of twisted pear's LCBPS, I don't know what changes i can make, but a super regulator may be a little overboard. I don't really know about these things, I'll take a serous look later when i have time.
What about classification? I know AMB sells GR, should I buy about 100 of each (Class Y)? and just match these transistors of the same batch? or should I get some more GRs?
Thank!
What about classification? I know AMB sells GR, should I buy about 100 of each (Class Y)? and just match these transistors of the same batch? or should I get some more GRs?
Thank!
Twisted pear's LCBPS looks ok for basic regulated supply, but once your headamp is up and running, you know what is a weak point. Also, they should design in RCRC instead CRC filter, you can improve things if you can mount another resistor in series with R1/R2 between A1/A2 diode bridge positive/negative output and C3/C4 and cut the traces wich connect them. New added resistor should be low in value, say 1-2R2, R1/R2 6R8-10R.
Also, you can improve things if you replace D1/D2 with 1N4148 (or 1N914) in series with 10R resistor, orientation for 1N4148 is the same as for 1N4002.
As for transistors classification, you sould look for highest Hfe, so try to find BL devices of the same batch. But, if you can't find them, GR devises will do.
Also, you can improve things if you replace D1/D2 with 1N4148 (or 1N914) in series with 10R resistor, orientation for 1N4148 is the same as for 1N4002.
As for transistors classification, you sould look for highest Hfe, so try to find BL devices of the same batch. But, if you can't find them, GR devises will do.
I've just open the BOM for Twisted pear's LCBPS and I can tell you they have design in a fatal mistake...output capacitors C9/C10 should not be very low impedance type, LM317/LM337 will resonate, so replace them with good quality, computer grade ones, nothing fancy. Or keep the Panasonic FM types but place in series with them 0R1/0,25W resistor.
I'll concentrate building a schematic for the PSU tomorrow, Should i add anything after the LM3x7 regulators to improve regulation? It's hard to find any circuits that provide measurements.
I'll make a PCB layout of a filter before the transformer with a 1Meg resistor, a 0.22uF cap, and a 1.9mH 4amp Common Mode Choke. I believe this will add stability and noise reduction, possibly eliminate the problem of my computers PSU.
As for parts ordering, I'll order 100each of the 2SA1015/2SC1815 that makes 200 transistors. Should get some perfect matches.
As for boards from AMB, I'll order the Dynalo board and possibly a[FONT=arial,helvetica] ε27?[/FONT] Is there anything else i should be ordering just to make sure everything goes smooth?
I'll comprise the pulmonary mouser order (hopefully) tomorrow for all the parts i need for tomorrow.
I'll make a PCB layout of a filter before the transformer with a 1Meg resistor, a 0.22uF cap, and a 1.9mH 4amp Common Mode Choke. I believe this will add stability and noise reduction, possibly eliminate the problem of my computers PSU.
As for parts ordering, I'll order 100each of the 2SA1015/2SC1815 that makes 200 transistors. Should get some perfect matches.
As for boards from AMB, I'll order the Dynalo board and possibly a[FONT=arial,helvetica] ε27?[/FONT] Is there anything else i should be ordering just to make sure everything goes smooth?
I'll comprise the pulmonary mouser order (hopefully) tomorrow for all the parts i need for tomorrow.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.