Hi there.
I was trying to figure out how to pass by the opamps of my Hypex UCD to improve the SQ of the amps, when another idea hit me.
I am running a TA2021B on my main system.
As we all know , there is an inverting opamp sitting on the input of the Tripath amps.
Gain is V0/V1=-12*RF/FI
Now what would happen if I open the feedback loop and connect my input signal to the feedback pin without connecting it to the opamp input?
I guess I'd end up at an gain of 12!? Would this be working? What happens to the impedance? Any other issues I might have to look at?
Cheers
I was trying to figure out how to pass by the opamps of my Hypex UCD to improve the SQ of the amps, when another idea hit me.
I am running a TA2021B on my main system.
As we all know , there is an inverting opamp sitting on the input of the Tripath amps.
Gain is V0/V1=-12*RF/FI
Now what would happen if I open the feedback loop and connect my input signal to the feedback pin without connecting it to the opamp input?
I guess I'd end up at an gain of 12!? Would this be working? What happens to the impedance? Any other issues I might have to look at?
Cheers
Hi folks.
I'd like to come back on that one, since I am gonna touch my amp soon.
I'd like to rephrase my question:
How about shortening RF? This should give a gain of 1 on the opamp input stage - correct? This would be basic opamp behavior.
But - according to below formula taken from the datasheet the overall gain becomes 0 if I shorten RF.
vO/vI=-12*Rf/Ri
TA2021B datasheet pdf
I'd really appreciate if somebody could give advise, what's gonna happen if I shorten Rf. Otherwise I'll potentially fry the amp.
THX a lot.
I'd like to come back on that one, since I am gonna touch my amp soon.
I'd like to rephrase my question:
How about shortening RF? This should give a gain of 1 on the opamp input stage - correct? This would be basic opamp behavior.
But - according to below formula taken from the datasheet the overall gain becomes 0 if I shorten RF.
vO/vI=-12*Rf/Ri
TA2021B datasheet pdf
I'd really appreciate if somebody could give advise, what's gonna happen if I shorten Rf. Otherwise I'll potentially fry the amp.
THX a lot.
Great.
BTW: You got a small mistake in your drawing. RF is feeding the inverting input
on the 2nd opamp!
Perhaps some words about the underlying story.
I am running a 1.7V DAC output right into the amp firing 98db/SPL speakers.
This forces me to apply 12-18db attenuation - digitally on the PC side. This is of course not acceptable for a 16bit signal. (I'm still running a 1543 based DAC) I would like to get the gain as much down as possible.
Perhaps a nice side effect could be that I'll improve the performance of the amp if I can get rid of RF. Perhaps a nice tweak also for others running sensitive speakers.
To go a step further the first idea was to leave out the entire input stage and to feed the signal through RI ( disconnected from the opamp input ) right into the feedback output. Perhaps it is also worth to test this approach.
Feel free to comment.
I am running a 1.7V DAC output right into the amp firing 98db/SPL speakers.
This forces me to apply 12-18db attenuation - digitally on the PC side. This is of course not acceptable for a 16bit signal. (I'm still running a 1543 based DAC) I would like to get the gain as much down as possible.
Perhaps a nice side effect could be that I'll improve the performance of the amp if I can get rid of RF. Perhaps a nice tweak also for others running sensitive speakers.
To go a step further the first idea was to leave out the entire input stage and to feed the signal through RI ( disconnected from the opamp input ) right into the feedback output. Perhaps it is also worth to test this approach.
Feel free to comment.
Last edited:
That's straight out of the TC2001 datasheet. Tripaths fault, not mine.
When Rf is shorted, everything gets attenuated (signal AND noise). I was kind of excited until I shorted Ri as well. With Ri shorted the signal level jumps back up, but it doesn't sound right at all. All the bass is gone and the signal is a bit distorted.
Anyway, all the poking and prodding of the input pins didn't kill anything, so you should be safe to experiment on your amp.
If there is something else you want me to try while I have all this mess out let me know. I'll be putting all back away shortly.
THX a lot real good news.
So. Shortening RF is working. And you even got excited. Cool.
Can you figure out what gain we're looking at?
Why did you shorten RI?
I was thinking of cutting RI off from the inverting input to feed the signal straight in the RF output ( RF stays shortened). This would keep the opamp completely out of the loop.
Cheers
So. Shortening RF is working. And you even got excited. Cool.
Can you figure out what gain we're looking at?
Why did you shorten RI?
I was thinking of cutting RI off from the inverting input to feed the signal straight in the RF output ( RF stays shortened). This would keep the opamp completely out of the loop.
Cheers
Because with just Rf shorted the output level was very low, even with an extremely high input signal.
Ehhh..... I'm not up for math at the moment. I'll just post some voltage readings and you can glean from it what you wish.
These output readings are with a 0.3Vrms input signal (sine @200Hz)
Rf and Ri normal = 2.9Vrms
Rf shorted = 0.04Vrms
Rf and Ri shorted = 2.2Vrms
An interesting thing; the 200Hz sine output sounds like a square wave when Rf and Ri are both shorted.
I dunno soundcheck, looks like you'll have to do some tests of your own. I think I've had enough fun for now.
I think they just forgot the - in the 2020 datasheet. It's obviously an inverting gain stage, so the 2021B datasheet is correct.
Yes, the outputs are inverted.
THX a lot for doing the testing. Very interesting.
0.04V that doesn't sound a lot.
Av/db=20log(0,04/0,3)=-17,5db
We're actually loosing 17db. That much I didn't expect. The formula seems "almost" right. Though it's obviously not.
I am wondering if my 2021B would exactly behave like your 2001. I need to check that out. The 2001 datasheet says clearly the overall gain is a product of the modulator gain and the inputstage gain. My guess would be that the modulator gain is fixed at 12 on the 2021B and yours is variable according to the formula.
In my particular use case a little higher input voltage would be required to make my case fly.
-17db is a bit too much I guess. However I'll give it a try.
BTW: Is there a recommendation in which range RI or RF should be dimensioned in absolute terms? Ri=20||30k||40k......??
Cheers and TX again.
0.04V that doesn't sound a lot.
Av/db=20log(0,04/0,3)=-17,5db
We're actually loosing 17db. That much I didn't expect. The formula seems "almost" right. Though it's obviously not.
I am wondering if my 2021B would exactly behave like your 2001. I need to check that out. The 2001 datasheet says clearly the overall gain is a product of the modulator gain and the inputstage gain. My guess would be that the modulator gain is fixed at 12 on the 2021B and yours is variable according to the formula.
In my particular use case a little higher input voltage would be required to make my case fly.
-17db is a bit too much I guess. However I'll give it a try.
BTW: Is there a recommendation in which range RI or RF should be dimensioned in absolute terms? Ri=20||30k||40k......??
Cheers and TX again.
The lower the better?
I had them a 1k/1k on one of my AMP32s to see if it would lower the background noise any. Can't remember if it did or not.
@theAnonymous1
I still don't give up yet. the whole thing is bugging me. Don't hope that I am bothering you too much.
What have been your amp Ri/Rf values and what is your source Ro.
I did some assumptions and simple calculations based on the assumption that your Ri=20k.
It seems that in case of Ri=2k66 and Rf=short we should end up at unity gain. (I'll write that up later on - as soon as it is proved that I am right here)
The damping factor should be high enough now ( >10/1 -- obviously depending on your Rosource, which I don't know yet) to avoid the distortions you experienced in your RF=RI=short case.
I calculated that your source might have seen an Riamp of 365R in your RF=RI=short case. This related poor damping factor could have been the reason for the distorted sound you were experiencing. But I am not really sure here.
If you're interested to run one more test -- I'd really like to see what's happening on the amp output if you set RI=2k66 and RF=short.
Cheers
I still don't give up yet. the whole thing is bugging me. Don't hope that I am bothering you too much.
What have been your amp Ri/Rf values and what is your source Ro.
I did some assumptions and simple calculations based on the assumption that your Ri=20k.
It seems that in case of Ri=2k66 and Rf=short we should end up at unity gain. (I'll write that up later on - as soon as it is proved that I am right here)
The damping factor should be high enough now ( >10/1 -- obviously depending on your Rosource, which I don't know yet) to avoid the distortions you experienced in your RF=RI=short case.
I calculated that your source might have seen an Riamp of 365R in your RF=RI=short case. This related poor damping factor could have been the reason for the distorted sound you were experiencing. But I am not really sure here.
If you're interested to run one more test -- I'd really like to see what's happening on the amp output if you set RI=2k66 and RF=short.
Cheers
Last edited:
Perhaps the input signal level to the modulator was just too small at that point.
If the amp output was 0.04V and assume an AV of the modulator of around 7.3, the modulator must have seen an input voltage of approx. 6mV. Which is pretty low I'd say. At unity gain we probably can expect 43mV at that stage.
If your PGA comes with Ro=50R values down to 500R on Riamp should be manageable.
Using a Ri=1k might be a good value for operating the amp in that RF-short mode. It should deliver then 0.8V Uamp out which would equal 8,5db gain at 0,3V Ui.
This 8.5db gain would be pretty close to what I need in my use-case.
To generate my preferred SPLmax = 98db/SPL P=1W || Voamp=4Vrms @ 16R and Vodac=1,6Vrms @ 0dbfs I'd need a gain/db=20log(Voamp/Vodac)=7,95db.
So far the (my) theory.
Of course 8db gain should be achievable also at RF/RI=4k12/20k.
Cheers
OK....
With the same input signal as before (0.3Vrms sine @200Hz)....
Ri = 1k
Rf = 0R (short)
Vo = ~0.75Vrms
Another observation; the normal Ri/Rf values will produce a ~6Vrms output before there is audible clipping. With Ri/Rf = 1k/0R audible clipping is reached at only ~1.9Vrms.
There is something else very strange going on that I have to think of a way to explain.
With the same input signal as before (0.3Vrms sine @200Hz)....
Ri = 1k
Rf = 0R (short)
Vo = ~0.75Vrms
Another observation; the normal Ri/Rf values will produce a ~6Vrms output before there is audible clipping. With Ri/Rf = 1k/0R audible clipping is reached at only ~1.9Vrms.
There is something else very strange going on that I have to think of a way to explain.
There is something else very strange going on that I have to think of a way to explain.
If I match the SPL of the drivers by ear from one normal channel and one modified channel.....
Channel 1
(Ri/Rf = 20k/20k)
Vo = 2.9V
Channel 2 (Ri is a trim pot)
(Ri/Rf = 395R/0R)
Vo = 1.47
Please explain how both drivers can produce the same SPL, but the unmodified channel has twice the voltage?
It's times like this I wish I had a scope.
Last edited:
At Ri=1k Uo=0.75V that's what I predicted. wasn't too difficult!
It needs 2 times the voltage to go 3db up on the electrical power side. That doesn't mean necessarily that the SPL goes up also 3db. There should be a slightly audible difference though. I guess it'll be hard to distinguish without measuring it.
BUT: The clipping at Uoamp=1.9Vrms is not really nice. What's that supposed to mean? I need to think about that one.
Did you try to feed the signal right into the RF output to avoid the opamp totally? If not - never mind. I think we gotta pretty good picture what's possible with the device. I am not aware that anybody else tried these "tweaks".
Before I forget. How is the sound up to 1.9VRMS? Still distorted?
Anyhow . THX a lot for your support so far. I really appreciate it.
wasn't too difficult! It needs 2 times the voltage to go 3db up on the electrical power side. That doesn't mean necessarily that the SPL goes up also 3db. There should be a slightly audible difference though. I guess it'll be hard to distinguish without measuring it.
BUT: The clipping at Uoamp=1.9Vrms is not really nice. What's that supposed to mean? I need to think about that one.
Did you try to feed the signal right into the RF output to avoid the opamp totally? If not - never mind. I think we gotta pretty good picture what's possible with the device. I am not aware that anybody else tried these "tweaks".
Before I forget. How is the sound up to 1.9VRMS? Still distorted?
Anyhow . THX a lot for your support so far. I really appreciate it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.