Hi,
I have build a gainclone with the LM3875 with the scematic of Fedde (Chill-amp, not inverted, included zobel)
I designed my own board with 2 amps and the PSU on the same board.
When connecting it, I'm measuring a DC offset of -85 mV on the output (input shorted and output R 10 Ohms). When I switch the amp off, the DC voltage rises to 1.5 V and decreasing to 0 V. You also hear a "pop" from the speakers when turning off.
What can I do to get rid of the DC voltage? Resistor between input cap and pin 7 or or include Ci??? Oscillation...?
The amp is pretty quiet whit no input connected and it plays fine, but I think -85 mV is to much!!
Here a picture of my PCB
I hope someone can give me some good advise!!
D Siesling
I have build a gainclone with the LM3875 with the scematic of Fedde (Chill-amp, not inverted, included zobel)
I designed my own board with 2 amps and the PSU on the same board.
When connecting it, I'm measuring a DC offset of -85 mV on the output (input shorted and output R 10 Ohms). When I switch the amp off, the DC voltage rises to 1.5 V and decreasing to 0 V. You also hear a "pop" from the speakers when turning off.
What can I do to get rid of the DC voltage? Resistor between input cap and pin 7 or or include Ci??? Oscillation...?
The amp is pretty quiet whit no input connected and it plays fine, but I think -85 mV is to much!!
Here a picture of my PCB
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
I hope someone can give me some good advise!!
D Siesling
Dirk(?),
The DC path to gnd from the non-inv input is 22k, from the inverting inut it is 680ohms//22k = 659.6 ohms. There is a DC bias current at the inputs, and because of the unequal resistances, this generates a DC difference voltage at the inputs which is amplified and appears at the output. Make the non-inv input 10K (your preamp can handle this unless it is a very high output impedance tube amp), replace the 680 ohms by 10 k also and change the feedback resistor to 180k. This will make the impedances almost equal, and makes the gain lower so your volume control will be less nervous.
If you really want to balance the DC paths, make the non-inverting input R to gnd up of 10k//180k.
Note that the bias current are never exactly equal so even with equal resistors you may still have some offset, so the extra accuracy probably isn't worth it.
Jan Didden
PS You can post directly to the forum, of course. No need for emails.
The DC path to gnd from the non-inv input is 22k, from the inverting inut it is 680ohms//22k = 659.6 ohms. There is a DC bias current at the inputs, and because of the unequal resistances, this generates a DC difference voltage at the inputs which is amplified and appears at the output. Make the non-inv input 10K (your preamp can handle this unless it is a very high output impedance tube amp), replace the 680 ohms by 10 k also and change the feedback resistor to 180k. This will make the impedances almost equal, and makes the gain lower so your volume control will be less nervous.
If you really want to balance the DC paths, make the non-inverting input R to gnd up of 10k//180k.
Note that the bias current are never exactly equal so even with equal resistors you may still have some offset, so the extra accuracy probably isn't worth it.
Jan Didden
PS You can post directly to the forum, of course. No need for emails.
The circuit from Fede's site www.fedde.tk
mine is almost the same, only I use 4.7u in stead of 2 in parallel.
D Siesling
mine is almost the same, only I use 4.7u in stead of 2 in parallel.
An externally hosted image should be here but it was not working when we last tested it.
D Siesling
Yeah, I agree with tube_dude.
For more information look at datasheet for LM3875 at www.national.com/pf/LM/LM3875.html.:D
For more information look at datasheet for LM3875 at www.national.com/pf/LM/LM3875.html.:D
I have used DC servos on both my Gainclones and one advantage is a totally silent power-on and power-off. Some of the disturbance is eliminated by the servo because it works under 12 volts which is the limit for LM3875/3886.durxter said:
A simple design may have drawbacks and the one you describe is one of them if it really bugs you.
85mV isn't extremely much and will probably not have any influence on the performance of your speakers.durxter said:
When your homepage is working (nothing there yet?) you are very welcome to join my webring for Gainclones
http://C.webring.com/hub?ring=gaincloneamplifi
durxter said:
Durk,
Let me ask you a question in return: If you have no idea about the influence of the cap or what type/value cap you need (which is not meant as derogatory, we all learned the hard way), what was your reason to decide to pick this solution over the other ones offered?
Jan Didden
Jan,
I' ve tried your solution first actually!! and it works, only 5,8 mV!! and it looks like the nasty pop in my speaker has gone to.
I first thought adding the cap was the most logic thing to do because it's in the datasheet. But I didn't want extra caps in the circuit when not really necessary. Thats wy I chose your option after all. It was also much easier to do on my PCB, no room for a extra cap.
Thank you for this solution, it works great!!
D Siesling
I' ve tried your solution first actually!! and it works, only 5,8 mV!! and it looks like the nasty pop in my speaker has gone to.
I first thought adding the cap was the most logic thing to do because it's in the datasheet. But I didn't want extra caps in the circuit when not really necessary. Thats wy I chose your option after all. It was also much easier to do on my PCB, no room for a extra cap.
Thank you for this solution, it works great!!
D Siesling
Durk,
I wasn't trying to force my suggestion on you, just wondering about your reasons.
Glad it works. You probably still have *some* pops, but hopefully you can live with it.
Now, take the money you saved from that expensive cap, buy I bottle of wine and call your (girl) friend!
Happy new year, and happy listening!
Jan Didden
I wasn't trying to force my suggestion on you, just wondering about your reasons.
Glad it works. You probably still have *some* pops, but hopefully you can live with it.
Now, take the money you saved from that expensive cap, buy I bottle of wine and call your (girl) friend!
Happy new year, and happy listening!
Jan Didden
listening impressions
Hi everybody,
Today a listened seriously to my GC the first time. I compared it with my present amp ( The AXL- class-A amp from Elektor 1985, based on the "mini crescendo") with Fostex DIY TQWP speakers connected.
I was really impressed!! The GC is much more detailed en brighter than the AXL. Also there is al little more bass, the TQWP's need that with their 9cm drivers. So I'm very pleased with it. Now I only have to build a nice alu case for it.......
D Siesling
Hi everybody,
Today a listened seriously to my GC the first time. I compared it with my present amp ( The AXL- class-A amp from Elektor 1985, based on the "mini crescendo") with Fostex DIY TQWP speakers connected.
I was really impressed!! The GC is much more detailed en brighter than the AXL. Also there is al little more bass, the TQWP's need that with their 9cm drivers. So I'm very pleased with it. Now I only have to build a nice alu case for it.......
D Siesling
If you have 85 mV DC voltage on the output this is about 1mW dissipation in the speaker coil. I am not worried about that at all. It could be even benificial, to keep it a half a degree warmer 
I would be more concerned about the sonic differences after changing the circuit. Though differences can be positive naturally...
BTW: as you can see, I found some time to update the schematic. I really recommend the zobel values in the new schematic above the previous.
Fedde
I would be more concerned about the sonic differences after changing the circuit. Though differences can be positive naturally...
BTW: as you can see, I found some time to update the schematic. I really recommend the zobel values in the new schematic above the previous.
Fedde
Fedde,
What sonic difference would it make with these resistor values in stead of the original? And what is the advantage of de new zobel values?
I am not really satisfied with the sound actually. I am using my MOSFET AXL amp again at the moment. The gainclone has much sharper high, I am missing the warmth of the MOSFEt amp. Maybe it is just a bad combi with my fullrange speakers.
Durk
What sonic difference would it make with these resistor values in stead of the original? And what is the advantage of de new zobel values?
I am not really satisfied with the sound actually. I am using my MOSFET AXL amp again at the moment. The gainclone has much sharper high, I am missing the warmth of the MOSFEt amp. Maybe it is just a bad combi with my fullrange speakers.
Durk
The new zobels values improve potentially the stability. If you have oscillation problems, chances are high they will be solved with this zobel. And then your sound gets much warmer. A gainclone should certainly not sound bright. But a gainclone is fast and gives a different sound as a mosfet amp. A mosfet amp could sweeten the sound a bit. Also, take care of a good source, as a gainclone is revealing.
Ciao,
Fedde
Ciao,
Fedde
Well, actually everything matters in a gainclone. I can't comment on the sonic difference of your resistor configuration. If the sound is really bright and unnatural, it is usually oscillation. Your PCB is quite large, this also doesn't really help. Did you mount the feedback resistor close to the chip? Just first try the zobel and please let us know the result. As second experiment your could try to replace the 4.7 uF cap with a 100-200 ohm resistor (but try only with a source with no DC offset!). It makes the amp more vulnerable, so watch out. But it improves the sound quality further.
In general, I would advise to experiment a little to get some feeling with the components in your amp.
Good luck,
Fedde
In general, I would advise to experiment a little to get some feeling with the components in your amp.
Good luck,
Fedde
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.