Need help getting this basic DIY amp up and running (DC offset Problems) - diyAudio
Go Back   Home > Forums > Amplifiers > Solid State

Solid State Talk all about solid state amplification.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 22nd December 2014, 03:36 AM   #1
diyAudio Member
 
Join Date: Jun 2010
Location: Williamsport, PA
Default Need help getting this basic DIY amp up and running (DC offset Problems)

Hi guys,

I am an electronics engineer that usually deals with non audio class d amplifiers. I just recently got interested in linear audio amplifiers and was intrigued by the how low cost a basic linear amp can be to thrown together. So after a bunch of reading through the forum and the art of electronics, I put together this amp which I spiced in LTspice and all seems well. It has pretty much all the basic parts that seem to be pretty common so I think you guys will be able to help. Initial start-up, It has high frequency oscillations so I increase the miller compensation cap. Surprisingly it needs to be much higher than ltspice says to become stable. The second problem and currently the biggest problem is that there is a 7.3V positive dc offset on the output. I tried messing with most stuff like the current sources by probing resistors in parallel. I tried finger probing around to find a location that finger resistance and capacitance would tweak the circuit response. Everything resulted in an identical output except altering the 3k feedback resistor. When I decrease the feedback resistor the AC gain decreases as expected but the DC offset also decreases, it seems by the same proportional amount. The AC gain is also completely wrong. It is supposed to be around 30 and it looks like the gain is around 2. I built this on a whim so it is powered by +-12v 6A laptop style supplies. I also intended on using 2n5551 NPN transistors but couldn't find them for the build so they are all 2n3904s which appears ok with low rails. I also have 2sd2390 and 2SB1560 on the final output stage. These are all what I had available to build with without ordering anything. I have no intentions on getting prefect hifi on this build I just want to get it working to play with. I probed the the -diff input transistor and it looks like a perfected 0VDC sinewave so interestingly it looks like the feedback thinks its working. DC on the output side of the 3k and no DC on the BJT side. Looking at the scope screenshot the yellow trace is the output and the magenta trace is the input. Also attached is the simulation I designed from and pics of the build. I think some of you guys will be able to quickly pick out my problem. I am looking forward to any thoughts. Thanks.
Attached Images
File Type: jpg 1221141956.jpg (390.9 KB, 198 views)
File Type: jpg 1221141956a.jpg (351.7 KB, 191 views)
File Type: jpg 1221141957.jpg (621.9 KB, 204 views)
File Type: png tek00310.png (22.5 KB, 192 views)
Attached Files
File Type: asc DIYamp.asc (4.8 KB, 33 views)
  Reply With Quote
Old 22nd December 2014, 10:45 AM   #2
diyAudio Member
 
jerluwoo's Avatar
 
Join Date: Feb 2006
Location: Ashland,Ky U.S
Try adding a capacitor between your 90ohm ? feedback resistor and ground. Having it grounded directly will cause large offsets unless the ltp transistors are perfectly matched. Also the resistor on the input side from base to ground should be the same value, can,t see it from the pic since it,s standing up. I would go with 10k's for this one and the feedback one going to output, 3k is a little low since this will define input impedance, and 560 for the smaller feedback resistor with 100uF capacitor to ground.

Last edited by jerluwoo; 22nd December 2014 at 10:54 AM.
  Reply With Quote
Old 22nd December 2014, 03:49 PM   #3
diyAudio Member
 
Join Date: Jun 2011
tl;dr

Unformatted paragraphs of such length are very wearing on the reader.
__________________
Now I am Become Death, the Destroyer of Words
  Reply With Quote
Old 23rd December 2014, 03:18 AM   #4
diyAudio Member
 
Join Date: Jun 2010
Location: Williamsport, PA
Jerluwoo,

Thanks for the tip. I do recall seeing that cap on other ac coupled amps. I gave it a try with mixed results. I put in a 470uf in series with the 100r to ground and it did seem to correct the dc as the cap charged. So at power on it had the 7v dc offset and then charged down toward 0v. The problem is once the sine wave crossed 0v it clipped the bottom of the wave. So that kinda bandaids the problem. But there seems to be a bigger issue.

Other updates and tests-

Added 500r trimpot to the emitters of the ltp with the wiper going to the ccs to the neg rail. I was able to get +-2v of dc offset adjustment but not enough.

I added a 4700uf cap to the output and so I could connect up a speaker. It seemed to drive them great. I was able to get ~25w into the 1.7ohm paralleled speakers before clipping. Though I think when everything is working right I think I should be able to get closer to 40w.

Switching to my phone as a signal source, it drove fine other than the fact that the phone can't drive more than like 200mv pk so it didn't get very loud with the phone all the way up. So that brought me back to the gain problem. It still only has a ac gain of 2 when it should be 30. I increased the negative feedback resistor to 10k from 3k so theoretical gain of 100. It resulted with the dc offset being completely in the positive rail and I had to put a negative offset on the function generator to get the wave into driving range. It did not increase the ac gain much at all.

I was very impressed with the speed of the amp . I tested square and triangle waves up to 1mhz and they looked Damn good with only some slew distortion and some hf hash in which might be coming from the test equipment because the hash was on the input signal from the agilent function generator. The input waves are blue and the outputs are yellow. The out is ac coupled on the scope. This was a low voltage test so the slews will probably get worse at high voltage. Way faster than audio needs but I might drive some piezo transducers with it at work.

Counter culture,

I will get that first post cleaned up. I spewed all my thoughts through my phone. Which is more difficult to format. Sorry for these first post lengths also I'm just trying to get enough info on here to get the correct help.

Im hoping that I get get some more advice and help. I'm eager to get a second one working to have stereo. Thanks in advance.
Attached Images
File Type: jpg 1222141816.jpg (991.0 KB, 127 views)
File Type: jpg 1222141813a.jpg (896.3 KB, 17 views)
  Reply With Quote
Old 23rd December 2014, 05:53 AM   #5
blu_glo is online now blu_glo  England
diyAudio Member
 
blu_glo's Avatar
 
Join Date: Aug 2005
Location: London, England
My tuppence worth -
1 - that D45H11 (?) PNP transistor is no compliment of the 2N3055 that I know of - try MJ2955.
2 - You have no emitter degeneration resistors nor base resistors in output stage. Suggest you start off with 0.47 ohms in series with each output transistor emitter and circa 100 ohms in series with their bases. You can then reduce to 0.22R min recommended emitter resistors and a token base resistor that helps with thermal stability...

attached is modified schematic.... some component values are up for tweaking!

Don't forget with DC coupled input this will pass any input DC offset on to the output!

Last edited by blu_glo; 23rd December 2014 at 06:13 AM.
  Reply With Quote
Old 23rd December 2014, 10:24 AM   #6
diyAudio Member
 
jerluwoo's Avatar
 
Join Date: Feb 2006
Location: Ashland,Ky U.S
A schematic of exactly what you have built would be a big help. I,m just guessing by what I could see in the pics.
  Reply With Quote
Old 24th December 2014, 12:37 AM   #7
diyAudio Member
 
Join Date: Jun 2010
Location: Williamsport, PA
Blu_Glo,

1 - These were the only 10A transistors in my ltspice library. I am not really using them in the real circuit. I am using sanken audio darlingtons 2SD2390 and 2SB1560 of which I don't have spice models to.

2 - I'm under the impression that the 100ohm base resistors are only necessary if there is a emitter follower frequency stability issue or to linearize vbe differences when paralleling output devices. I also assumed since that I was using a single output stage that the emitter resistor wasn't necessary because current sharing wasn't needed like in a parallel output setup. I do think I should put them in to stabilize the bias current.

There are some questions about some of your tweaks that I don't understand.

1. Can you explain C6 and R20? I have never seen them used that way. C6 breaks the necessary connection point for R9 which breaks the VBE multiplier. There is 6V of bias in your version causing 2A of bias current in the output stage.

2. What was wrong with the 4.7k resistors in the current sources? You have changed them to 10k but I don't see the reason.

3. The 47uf in the feedback path, I will probably add that eventually because it does fix offset problems but as I said in the previous post, it band-aids and covers up the real circuit problems. So once I get the offset in an acceptable range ill put it in.

4. Your C5 5.6pf cap I have not added because the frequency response is currently stable without it and I want to see the true part responses before compensating for safety.

5. interestingly the spice I uploaded didn't have any functionality problems but I ran yours and the output clips at 3v. I removed C6 and it fixed the clipping.

6. The output base resistors also limit the minimum load impedance. In the simulation the 100R resistors prevent driving below 7 ohms because the transistors cant get enough drive current. My real sanken output transistors have a really high hfe (datasheet say 20000) so the 100 ohms should work in my bench circuit but im not sure they are necessary. I think ill try them as a troubleshooting step though.


Jerluwoo,

attached is the circuit with all my real parts but the output transistors. ASC and PIC.

I realize 2n3904 transistors parameters are way off from part to part but I was expecting 100's of mv offset not 7V. I think the problem is in the feedback path because the gain is wrong too and the offset changes with altering the feedback resistor. Can someone chime in if they have seen huge offsets from unmatched transistors?

Does anyone know if too high of hfe in the output stage causes problems. From the emitter followers i've done in the past they just seem to follow even if there are matching issues then problems arise in the VAS because of unequal loading. I can imagine 20000 HFE can cause the signal to go unstable if not careful.

Thanks guys. We should be able to get it working soon.
Attached Images
File Type: png AsBuiltOnBench.png (57.2 KB, 65 views)
Attached Files
File Type: asc audio_amp.asc (5.7 KB, 4 views)
  Reply With Quote
Old 24th December 2014, 04:51 AM   #8
CBS240 is offline CBS240  United States
diyAudio Member
 
CBS240's Avatar
 
Join Date: Aug 2005
Location: K-town
No DC blocking cap? That is, a cap of appropriate value in series with R8 and sGND. Another option is to use DC servo to bias the DC input.
__________________
All the trouble I've ever been in started out as fun......
  Reply With Quote
Old 24th December 2014, 06:29 AM   #9
diyAudio Member
 
jerluwoo's Avatar
 
Join Date: Feb 2006
Location: Ashland,Ky U.S
OK a bit of a mash up of unequal items. R10 and R11 should be the same value, make them both 10k. Add capacitor to the input in front of R5, 10uF. R5 should be 1k and C2 1nF. Change R8 to 560 ohms and add a capacitor between it and ground, at least 100uF. Make R6 22-47 ohms. Q6 should be 2n3906. Add .1 to .22 ohm resistors to the emitters of the output transistors. Cross your finger and power it up!

Last edited by jerluwoo; 24th December 2014 at 06:33 AM.
  Reply With Quote
Old 24th December 2014, 07:24 AM   #10
blu_glo is online now blu_glo  England
diyAudio Member
 
blu_glo's Avatar
 
Join Date: Aug 2005
Location: London, England
- These were the only 10A transistors in my ltspice library. I am not really using them in the real circuit. I am using sanken audio darlingtons 2SD2390 and 2SB1560 of which I don't have spice models to.

1 OK that's what I assumed...

2 - I'm under the impression that the 100ohm base resistors are only necessary if there is a emitter follower frequency stability issue or to linearize vbe differences when paralleling output devices. I also assumed since that I was using a single output stage that the emitter resistor wasn't necessary because current sharing wasn't needed like in a parallel output setup. I do think I should put them in to stabilize the bias current.

To a degree they help in single output pairs to stabilise the IQ vs temperature. It gives the base current something to allow small voltage variation to be swamped...

There are some questions about some of your tweaks that I don't understand.

1. Can you explain C6 and R20? I have never seen them used that way. C6 breaks the necessary connection point for R9 which breaks the VBE multiplier. There is 6V of bias in your version causing 2A of bias current in the output stage.

someone had a crash course in operating LT Spice at 4:30 in the morning to answer your question!!C6 was not intended to break R9 connection - I got in a bit of a "mucking fuddle" attached is the corrected diagram! C6 and R20 help reduce Cdom distortion at higher frequencies if you are careful thd at 1K and 20K can be pretty much the same. I used LTSpice to correct your schematic only, and have just now run many simulations and got to see what it's about.

2. What was wrong with the 4.7k resistors in the current sources? You have changed them to 10k but I don't see the reason.

I doubled them as I took the ground ends to the other supply. this helps keep the amp running correctly at lower voltages and reduce switch on thump.... also prevents DC from flowing in your input circuit ground which I've found helps reduce THD somewhat

3. The 47uf in the feedback path, I will probably add that eventually because it does fix offset problems but as I said in the previous post, it band-aids and covers up the real circuit problems. So once I get the offset in an acceptable range ill put it in.

I'd make that capacitor a non polarised electrolytic as well, I don't encourage the use of polarised in this position.
The original spice circuit was showing dc offset which I got rid of tweaking R1, R2, R6



4. Your C5 5.6pf cap I have not added because the frequency response is currently stable without it and I want to see the true part responses before compensating for safety.

Sometimes you'll find if you take your amp past 100kHz or so the frequency response starts to rise. This is not good and the 5p6 (may need tweaking) helps roll it off sensibly.

5. interestingly the spice I uploaded didn't have any functionality problems but I ran yours and the output clips at 3v. I removed C6 and it fixed the clipping.

fixed - as mentioned above

6. The output base resistors also limit the minimum load impedance. In the simulation the 100R resistors prevent driving below 7 ohms because the transistors cant get enough drive current. My real sanken output transistors have a really high hfe (datasheet say 20000) so the 100 ohms should work in my bench circuit but im not sure they are necessary. I think ill try them as a troubleshooting step though.

Not "stop" so much as "reduce" surely, I did not see what you are expecting into 4 ohms? With darlingtons, you'd make those resistors several orders of magnitude higher, but again, open to tweaking of course taking into account your load requirements....
- Indeed if you're running darlingtons - you could possible get rid of Q5/R15/Q6.......



Last edited by blu_glo; 24th December 2014 at 07:47 AM.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
TDA7294 DC offset problems SCKentrol Chip Amps 39 18th May 2013 02:08 PM
Problems with DC offset Buhl Chip Amps 32 6th February 2006 07:51 AM
Gainclone with DC-offset problems seve Chip Amps 11 29th May 2004 12:01 AM
DC offset problems with leach amp wrl Solid State 62 3rd March 2004 01:28 AM


New To Site? Need Help?

All times are GMT. The time now is 07:30 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2015 DragonByte Technologies Ltd.
Copyright 1999-2015 diyAudio

Content Relevant URLs by vBSEO 3.3.2