Rarkov said:
Without you will end up with around 100mV DC Offset on the output. Since the feedback resistor is 1M, the "+" to ground needs to be the same, the input DC currents on both inputs are then balanced out and the DC Offset should be below 10mV typically. Occasionally the "+" resistor value needs to be tweaked for lowest Offset. The 0.1uF AC bypasses the 1M at audio frequencies, so the 1M is only seen at DC.
Joe R.
Nuuk said:
Very soon now with a bit of luck!
Hi Nuuk
Looking forward to it. Unlike some who don't build and yet come on here (or other threads) and use lazy intellectual criticism. Not that I am not anti-intellectual (like Hitler and other right wingers through history, even today) but lazy mental excercises are no substitute for elbow grease.
Not sure if you know, but there is a 3rd (or is it 4th?) option buffer, the "D" option that Pedja is now aware off, but one I'm yet to put on the table (I am toying with the idea). It is an SS substitute (not tube) that is used by us on the outputs of upgraded SACD players. It may not be tube but it is very good indeed.
Joe R.
Bloody hell Joe, I'm still trying to get the valve buffered IGC done and you and Pedja come up with more options
I have built two valve modules, one for either channel, one measures OK but on the other I get 20-30 volts after the output cap!
Both channels look exactly the same, I have swapped valves, checked the voltages, resistors and tried a differentoutput cap, tried the 'other side' of the valve and consulted PeterM. I can find no shorts and now can't think of anything else to try.Everything else is built, the input module, the amp modules and the case so as soon as I can sort out this second valve module, I hope to report back on the comparison with the OPA627 buffered version.
Well, I'm well confused but happier. After testing that dodgy valve module all of Friday and Saturday and getting the large amount of DC on the output what I tried or checked, this morning it reads 0mV. Have the hi-fi fairies been at work while I slept last night?
OK - on with testing the amp modules and then (hopefully) off we go with a full test.
OK - on with testing the amp modules and then (hopefully) off we go with a full test.
Nuuk said:
Bloody hell Joe, I'm still trying to get the valve buffered IGC done and you and Pedja come up with more options
In my own defense, I did say fairly early on that it didn't have to be a tube buffer, indeed encouraged you to try IC (and you picked a good one), but that tubes were simply my preference. The main thing was to establish that using a buffer was important and convince others.
I have a general hierarchy in mind, tubes first, then discrete SS whether fet or our 'SACD player buffer' (this is the "D" option which is not fet) and finally a good sonic grade IC. But even these can have sub catagories within them. Like my JLTi tube buffer isn't just CF and Pedja's fet is just a straight follower etc. So there are still variations to be tried and sampled (like a good bottle or two of vino).
I have posed the "D" option to Pedja and he has tried it out. He seems to like it (he isn't breaking a confidence as I am the one breaking the subject here). But is it better than any tube buffer? He's not been in a situation to confirm that. Whatever, it is a more complex circuit than I had in mind to post anyway. I will keep my powder dry for now... but am tempted in time.
We are all ears!
Joe R.
Keep those ears warmed up Joe!
I have just tested the amp modules and get 1.36 VDC on the output of both of them
I've checked the circuits, voltages etc and am at a total loss loss to explain the amount of offset. BTW, I tested them with a 47K resistor across the input.
Third pair of IGCs I have built and the first with any problems.
And just to make things worse, when I went to test the amp modules, I found my meter set to the AC setting so that probably explains why I got 0V reading on the output of that dodgy valve module. What a day!
Time to take a break.
I have just tested the amp modules and get 1.36 VDC on the output of both of them
I've checked the circuits, voltages etc and am at a total loss loss to explain the amount of offset. BTW, I tested them with a 47K resistor across the input.Third pair of IGCs I have built and the first with any problems.
And just to make things worse, when I went to test the amp modules, I found my meter set to the AC setting so that probably explains why I got 0V reading on the output of that dodgy valve module. What a day!
Time to take a break.
Hi Nuuk,
Just go with tubes and if Joe recommends that… Talking seriously, I never liked much ECC88, the idea of tube buffer never had much sense to me, but since I had a chance to check some Joe’s ideas and suggestions I don’t believe he went out lightly with the recommendation. Just get it to work, my guess is you won’t be sorry for the effort you put in it.
If it means something, I listened last days my BIGC with my EF86 based (anode follower) preamp. Not going into details, this has started me to think about and to take a first steps toward “my own” Tube BIGC.
As about the D option, it works great and no need to search for those K170 (I know it is painful for some). But it will be discovered exclusively by Joe (on the D day
).
Pedja
Just go with tubes and if Joe recommends that… Talking seriously, I never liked much ECC88, the idea of tube buffer never had much sense to me, but since I had a chance to check some Joe’s ideas and suggestions I don’t believe he went out lightly with the recommendation. Just get it to work, my guess is you won’t be sorry for the effort you put in it.
If it means something, I listened last days my BIGC with my EF86 based (anode follower) preamp. Not going into details, this has started me to think about and to take a first steps toward “my own” Tube BIGC.
As about the D option, it works great and no need to search for those K170 (I know it is painful for some). But it will be discovered exclusively by Joe (on the D day
).Pedja
Don't worry Pedja, I have every faith in Joe's suggestion for a valve buffer backed up by Peter M's experience with his.
Well, I have still got one dodgy valve buffer and 1.3 VDC on the outputs of the amps but to cheer myself up, I am posting a picture of the beast anyway.
As I just said to Pete M, if I can't get it working, I'll just power up the two blue LEDs under the valves and put it all on the sideboard as an ornament!
So my campaign against square-shaped boxes continues with this
Well, I have still got one dodgy valve buffer and 1.3 VDC on the outputs of the amps but to cheer myself up, I am posting a picture of the beast anyway.
As I just said to Pete M, if I can't get it working, I'll just power up the two blue LEDs under the valves and put it all on the sideboard as an ornament!
So my campaign against square-shaped boxes continues with this
An externally hosted image should be here but it was not working when we last tested it.
Thanks Moamps,
I was testing the amp modules on their own, ie not connected to the buffer. I just put a 47K resistor across the input and powered them up and both have 1.36 volts on the output (measured with a 10R resistor across the speaker terminals).
An externally hosted image should be here but it was not working when we last tested it.
The 47K is from the input to ground, and the 18K input resistor goes to the input of the LM3875 so I'm not sure how a cap could be in serial with both.
The answer is that apart from the decoupling caps, the only cap in the circuit is the 0.1uF which is in parallel with the 1 meg resistor from non-inverting input to ground.
In other words, the circuit is exactly as Joe's circuit AFTER the 3.3uF DC blocker BUT with the 47K from input (immediately before the 18K input resistor) to ground.
You can see the full circuit on Joe's site.
Ni Nuuk,
Moamps is probably right. It could be very likely you get the voltage you mentioned because you don’t have a cap at the inverting input’s path to ground in your testing circuit. Non-inverting input sees 1M, inverting input sees 47k//1M=44.89k DC resistance. (I’d rather expect to see negative voltage at the output in this case but I might be wrong here.) This input however should have an AC path to ground so I'd suggest you to try to put the cap in as shown below.
Pedja
Edit: Note that you are dealing with higher values resistors around the opamp now and this makes DC offset issues more "obvious".
Moamps is probably right. It could be very likely you get the voltage you mentioned because you don’t have a cap at the inverting input’s path to ground in your testing circuit. Non-inverting input sees 1M, inverting input sees 47k//1M=44.89k DC resistance. (I’d rather expect to see negative voltage at the output in this case but I might be wrong here.) This input however should have an AC path to ground so I'd suggest you to try to put the cap in as shown below.
Pedja
Edit: Note that you are dealing with higher values resistors around the opamp now and this makes DC offset issues more "obvious".
Attachments
fdegrove said:
Hmm, gut feeling, Frank...
I should look more for all the relevant info but, assuming no external injection of current (and there wasn’t any in this case) i.e. one separate opamp (and no multiple DC coupled stages), I always seen that the higher (relatively speaking) resistance from the LM3875's non-inverting input to ground means lower (or negative) offset at the output and this is related to the fact that this resistance always develops a negative voltage (NPN input). I guess it is the same with the majority of opamps.
Pedja
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.