Joe Rasmussen said:
Yes, I know.
Matching impedances is always good and is also a starting point for low DC.
So, just match impedances and hope for the best.😎
I will test the T-Network and valve buffer in a matter or 2~3 weeks, as I have some things to finish first.
So, I can not hide it:
Since converting to T-network, I enhanced the gain of my amp and I was fighting against hum...
Imho, this was a problem of the higher gain, not the T-network.
I could solve it now, by changing to a regulated PSU (LM338) for the powerstage (now with ultrafast diodes, before just a simple bridge rectifier).
But here my question:
Where do you connect the ground of the T-network? Signal ground or power ground?
Imho, signal ground is O.K.
Right?
Franz
Since converting to T-network, I enhanced the gain of my amp and I was fighting against hum...
Imho, this was a problem of the higher gain, not the T-network.
I could solve it now, by changing to a regulated PSU (LM338) for the powerstage (now with ultrafast diodes, before just a simple bridge rectifier).
But here my question:
Where do you connect the ground of the T-network? Signal ground or power ground?
Imho, signal ground is O.K.
Right?
Franz
Maybe, you have to reevaluate your results about big caps nearby the chips, with regulated PSU AND T-network (with OPA627 buffer and with tube buffer)!
The T-network is a very precise and fast feedback path and changes the situation.
As you could read in my posting before, I converted my VBIGC to a regulated PSU (2x10000uF). I used both diodes, recommended in the datasheet of the LM338, for higher caps after the regulators.
And I did not replace the 1000uF caps nearby the chip by lower values.
And the sound is great!
Franz
P.S.
I love it: you speak from valves, I speak from tubes. This remembers me, when I told to an english friend, I am building high end amps with tubes. Then, he asked me: "Franz, do they have valves, your amps?"
Franz G said:P.S.
I love it: you speak from valves, I speak from tubes. This remembers me, when I told to an english friend, I am building high end amps with tubes. Then, he asked me: "Franz, do they have valves, your amps?"
There are many ways to call the same thing.
My father was an EE (now retired), knows alot about valves and when I told him today what I was gonna do using the valve as a "buffer" on a power amp he said: oh, you're using it for "impedance matching".
"Buffer" doesn't exist in portuguese, there are other ways to call it.😀
A conventional TV screen is a valve.
Call it tube, it's ok.😎
Franz G said:
But you didn't try lower values.
Maby it can be even better?
Franz G said:
Cool.😎
carlosfm said:
I suppose if you wanted to be exact, then you would put 10K099 on (+) to ground. But 10K is still within 1% - so unless you want to be pedantic about it... ?
When you get around to do yours, let us know what DC offsets you get. The 10K based values are intended to keep minimum the effects of the input current offsets - that's what showing up on the output. It is not perfect, because if you are lucky to get good samples of 3875s, with low input offsets, then 10K will not give you lowest DC offset - but the 10K will be the best bet to get lowish offsets with the greatest variety of chips.
This will give the green light to newbies and not-so-new newbies, and if they've done a conventional feedback version, then there is no stopping them trying the T-network option and clearly get better results. It's about us giving a lead and make it accessible.
Joe R.
Joe Rasmussen said:
I will.
But I don't use the LM3875 for a long time, I prefer the LM3886.
But the plan is this:
I have a 2-channel board with LM1875s, and I will try on this one.
Doesn't even have a case yet, but it
ed-in on my main system for some time.Just figure out, this little amp with regulated PSU and zobel on the output drives very well my Epos 11 speakers (86~87db, 8 ohms)😱 .😎
First the T-Network on this great sounding little amp and then the tube buffer.😀
Carlos,
WRT the offset, LM1875 is not as tame as LM3875/3886. It may become more apparent with T feedback.
Pedja
WRT the offset, LM1875 is not as tame as LM3875/3886. It may become more apparent with T feedback.
Pedja
Pedja said:
I know, Pedja.
Normally it's a little higher.
My experience with this chip is that the tighter you make the layout the less DC you have.
Anyway, this should be a rule for every chip.
The first time i tried it I had around 50mv DC and used resistors on NI to ground to lower it.
On this very little 2-channel board I have around 10mv with NI directly to ground.
What's more, maby pure luck, DC is very similar on both channels.
😎
The interesting point here is measuring before and after implementing the T-network.
carlosfm said:
Let me explain what I mean with "tighter layout".
I made this board with two LM1875s and 2x100uf/35v caps on each chip.
Of course, 100uf caps are much smaller than 1000uf and even if I wanted now I couldn't fit 1000uf on-board on this amp.
The board is really small, and the 100uf caps are as near as possible to each chip's PSU pins.
I tested with my regulated variable LM338 test PSU.
But then I made more experiences with capacitance.
I like to try different approaches and I can tell you that even with an unregulated PSU the 100uf caps work very well.
So, with the amp board untouched (100uf) I tested with an unreg. PSU changing capacitors on the rectifier board.
The better result was with 2x3300uf caps.
Going higher is not good.
Going down to 2x2200uf is fine, but I feel that the sweet spot is between 2x2200 and 2x3300uf.
There are caps at values between those, but they are not so common and I didn't (and don't) have them to try.
Let's say the ideal could be 2x3000uf.
Easily done with 3x1000uf per rail.
In practice this is similar to having one unreg. PSU for two channels with the big 1500uf on the amp board, but the approach I tested had better results (and even lower DC😎 ).
What I've heard with this little amp (even before putting the zobel in) with unregulated PSU was certainly better than the results with my other LM1875 amp (2x1000uf per chip on board) with my Epos speakers.
My question is: is the Gaincard such a fine amp?
I've heard it with big full-range horn speakers and it sounded good, but it would not sound good at all with my speakers IMHO.
hmm, that doesn't seem to fit with my results. though i've only made 1 hi3gc. it seems the post i made last year about this same thing has gone away...
in anycase the entire board was 1.2" x 2.2", 2x1000uF, 2x 10uF, 2x 0.1uF bypassing.
even with the 70mV DC offset, the amp was great, albeit this was a woofer amp, so DC shouldn't affect it much. anyways.
in anycase the entire board was 1.2" x 2.2", 2x1000uF, 2x 10uF, 2x 0.1uF bypassing.
even with the 70mV DC offset, the amp was great, albeit this was a woofer amp, so DC shouldn't affect it much. anyways.
feedback resistors
It would be interesting to know the thoughts on the type of resistors for this T-network by those who have tried it.
I have a question regarding the feedback resistor in particular.
I know that we cannot ask "what is best" (or if we can, it may be too early to ask).
PD, for example, likes to use a Caddock MK132 in the feedback position, and I have listened to his recommendation. In our case here, we effectively have 2 feedback resistors -- in Joe's trials, 10k and 10k. I am assuming that any recommendation for a single resistor type would be valid for the 2 that we have here. BUT does the T-network affect the specific requirements for this position? Maybe the actual type of resistor is less relevant now, or different? Or maybe it all stays the same...
Thoughts, anyone?
It would be interesting to know the thoughts on the type of resistors for this T-network by those who have tried it.
I have a question regarding the feedback resistor in particular.
I know that we cannot ask "what is best" (or if we can, it may be too early to ask).
PD, for example, likes to use a Caddock MK132 in the feedback position, and I have listened to his recommendation. In our case here, we effectively have 2 feedback resistors -- in Joe's trials, 10k and 10k. I am assuming that any recommendation for a single resistor type would be valid for the 2 that we have here. BUT does the T-network affect the specific requirements for this position? Maybe the actual type of resistor is less relevant now, or different? Or maybe it all stays the same...
Thoughts, anyone?
More resistors now
If you're fiddling with resistor types and brands and listening to the amp on each resistor change you have work for some years.😀
If you're fiddling with resistor types and brands and listening to the amp on each resistor change you have work for some years.😀
Re: Re: More resistors now
I guess you're right...🙂
falcott said:Isn't that why we do this?😕 😀
I guess you're right...🙂
A word of warning may not go amiss here. The T-feedback experiment will usually involve soldering to the pins of the chip and then removing the resistor again later. We know how fragile the pins can be so extra care must be taken when doing this work.
That's the main reason that I have decided to do my T-network experiment with a new GC!
As regards resistor choice, the T-network makes this a much more complex affair as we have a combination of three resistors where there was one before. Keeping in mind the above caution, swapping resistors around too often isn't recommended.
Perhaps the 'safe' option is to use three different types of resistor!
That's the main reason that I have decided to do my T-network experiment with a new GC!
As regards resistor choice, the T-network makes this a much more complex affair as we have a combination of three resistors where there was one before. Keeping in mind the above caution, swapping resistors around too often isn't recommended.
Perhaps the 'safe' option is to use three different types of resistor!
Nuuk said:
As at least one nutritionist said: Each main meal should contain at least three vegestables.
Does the same apply to using resistors in amplifiers?
Joe R.
- Status
- Not open for further replies.