Good opamp for that1646

Help Support GroupDIY:

57sputnik

Active member
Joined
Jul 8, 2010
Messages
29
THAT1646 Output Noise vs. Source Impedance

1646_Noise_vs_Source.jpg


Output CMRR is not the only parameter that benefits from a low driving source impedance.

JRC end of life notice: Japan Radio NJM Linear End of Life Notice - Pro Audio Design Forum
LM4562/LME49720 EMI Problems: LME49720 DECT Cordless Phone EMI Susceptibility - Pro Audio Design Forum
 

abbey road d enfer

Well-known member
Moderator
Joined
Jan 22, 2008
Messages
13,480
Location
Marcelland
THAT1646 Output Noise vs. Source Impedance

1646_Noise_vs_Source.jpg


Output CMRR is not the only parameter that benefits from a low driving source impedance.
Is it a graph you produced yourself? What method?
It'd be interesting to know what alternative they propose...
It seems to confirm that what has been violently considered by "unbelievers" results of bad experimentation, is actually a fact. In the earlier posted thread.
The opponent claims that the circuit used to put in evidence the problem is not a viable circuit (86dB noise gain), but isn't it how über opamps' THD and other parameters are evaluated for publication?
 

57sputnik

Active member
Joined
Jul 8, 2010
Messages
29

57sputnik

Active member
Joined
Jul 8, 2010
Messages
29
Found the sim for THAT output common mode rejection vs. Rsource with an 18KΩ common mode impedance (THAT124X) load.

1646_Output_CMR_vs_Rs_Rcm_18K.jpg
 

abbey road d enfer

Well-known member
Moderator
Joined
Jan 22, 2008
Messages
13,480
Location
Marcelland
Found the sim for THAT output common mode rejection vs. Rsource with an 18KΩ common mode impedance (THAT124X) load.
Yes, it's similar to the CMRR sim I did with LTspice, but I didn't sim the noise behaviour, after having concluded that the damage to CMRR was enough to forget about it.
 

JohnRoberts

Well-known member
Staff member
Moderator
Joined
Nov 30, 2006
Messages
21,494
Location
Hickory, MS
The opponent claims that the circuit used to put in evidence the problem is not a viable circuit (86dB noise gain), but isn't it how über opamps' THD and other parameters are evaluated for publication?
I have long been critical of that method to characterize THD et al. Any IC designer worth a flip would design new ICs to shine under those specific test conditions. That said I appreciate the difficulty parsing out nonlinearities that deep in the dirt.

I can't imagine a prudent design using anywhere near that much noise gain.

JR
 

abbey road d enfer

Well-known member
Moderator
Joined
Jan 22, 2008
Messages
13,480
Location
Marcelland
I have long been critical of that method to characterize THD et al. Any IC designer worth a flip would design new ICs to shine under those specific test conditions. That said I appreciate the difficulty parsing out nonlinearities that deep in the dirt.
I agree. These accelerated methods are an ersatz, and it should be understood. Unfortunately datasheets are not too verbose about these methods. Caveat user.

I can't imagine a prudent design using anywhere near that much noise gain.
That's correct, and the guy who published his findings was clear about it. It is well known that FET opamps are less susceptible than bipolars and some bipolars are more equal than others in that respect. This guy received flak from people who criticized his method but did not offer a substitute. They just bitched him about bad implementation and lack of competence.
 

warpie

Well-known member
Joined
Feb 7, 2009
Messages
1,317
So, I breadboarded the whole thing and at first glance it seems to be working OK butI noticed that I get a difference of around 2.5mV between OUT+ and OUT-.

In fact, I'm reading the following voltages: OUT+ = -0.5mV and OUT- = +3mV.

Is this normal? The +/- voltages powering the 1646 also have around 2.5mV difference so this might be the reason(?) As I mentioned before I'm using resistors to drop the voltage going into the 1646. Should these be matched?

Also, shouldn't I get +mV at OUT+ and -mV at OUT- ?
 
Last edited:

warpie

Well-known member
Joined
Feb 7, 2009
Messages
1,317
Not really understanding. Are those DC Voltages readings ? And don't use resistors to drop voltage rails. The voltage drop will not be stable.

Yes this are the readings between GND and OUT+ and between GND and OUT-. Also, measuring with the DMM between OUT+ and OUT- shows around 2.5mV.

So, better to use regulators? I was just trying to save a bit on components, space and complexity...
 

Newmarket

Well-known member
Joined
Oct 10, 2016
Messages
838
Location
Brighton Sussex UK
Yes this are the readings between GND and OUT+ and between GND and OUT-. Also, measuring with the DMM between OUT+ and OUT- shows around 2.5mV.

So, better to use regulators? I was just trying to save a bit on components, space and complexity...

Offset voltages seem in line with the datasheet specs' and text.
Yes - regulators. For simplicity just rum the the opamp and the output IC from same voltage rails.
 

warpie

Well-known member
Joined
Feb 7, 2009
Messages
1,317
Would a 15V zener diode work here instead of a voltage regulator? I need to drop the voltage from +/-28VDC to +/-15VDC.
 

abbey road d enfer

Well-known member
Moderator
Joined
Jan 22, 2008
Messages
13,480
Location
Marcelland
Would a 15V zener diode work here instead of a voltage regulator? I need to drop the voltage from +/-28VDC to +/-15VDC.
Zeners can be noisy, and don't provide good regulation.
You should use LM317/337, not 7815/7915.
BTW, output offset will probably not improve with regulated rails, but some performance aspects may improve.
The primal cause of offset is the difference of Vbe (Vgs for FET) between the two input devices.
 

warpie

Well-known member
Joined
Feb 7, 2009
Messages
1,317
Zeners can be noisy, and don't provide good regulation.
You should use LM317/337, not 7815/7915.
BTW, output offset will probably not improve with regulated rails, but some performance aspects may improve.
The primal cause of offset is the difference of Vbe (Vgs for FET) between the two input devices.

Why LM317/337 instead of 7815/7915? I already use LM317/337 to produce +/-28VDC so I was thinking to use locally (close to the ICs) the 7815/7915 to drop the +/-28VDC to +/-15VDC. Not a good idea?

In fact, I had the impression that even the lower (0.1A) regulators would be sufficient to power two 5534s and two 1646s for a stereo unit, so the 1.5A 7815/7915 would be overkill.

In general I'm trying to understand what the benefits would be if I use regulators instead or resistors to drop the voltage. I mean practical benefits in the real world. I'm not too fuzzy about achieving that extra 0.0001% improvement in noise, thd, etc...

EDIT: I should mention again that the intended circuit is to be used to the zener limiter from fripholm which it's an old design and not a state of the art in terms of specs anyway so maybe some of the benefits will be masked anyway by the specs of the limiter itself.
 
Last edited:

Newmarket

Well-known member
Joined
Oct 10, 2016
Messages
838
Location
Brighton Sussex UK
tbh if you're using high spec audio semiconductor parts then why would you want to degrade the circuit with 'zener' regulation rather than a quiet and well established solution ???
Out of interest what value resistors are you thinking to use in conjunction with the zener diode ?
 

warpie

Well-known member
Joined
Feb 7, 2009
Messages
1,317
Out of interest what value resistors are you thinking to use in conjunction with the zener diode ?

I haven't thought about it. It was just a quick thought of using diodes instead of regulators.

I might use LM317/337 and call it a day but still I would like to know what would be the real benefits in using regulators. What improvements should I expect? Better noise figures? Better thd? Better CMR? Better everything? :)

Another concern I have is that the regulators will have to work hard to drop the voltage from 28VDC to 15VDC (i.e heat, big heatshinks, etc..)
 

Newmarket

Well-known member
Joined
Oct 10, 2016
Messages
838
Location
Brighton Sussex UK
Another concern I have is that the regulators will have to work hard to drop the voltage from 28VDC to 15VDC (i.e heat, big heatshinks, etc..)
Any Resistor/Zener combo' needs to dissipate the same power - c. 13V x ?mA. (operating/load current for the 1646 plus some 'Zener' Current).
If power dissipation in a regulator is an issue (I haven't looked at the current needed) then it's better to drop voltage on the way into the regulator eg via series resistors / diodes such that you retain a regulated voltage rail for the 1646.
 

Latest posts

Top