TA-7136P external frequency compensation in a Yamaha PM-430

[rjb5191]

Does anyone have a ballpark calculation for the rolloff caused by the 10 pf capacitor between pin 1 and 6 in the Ta-7136p gain stages. I wasn't paying attention and accidentally replaced one with a 22 pf c0g instead of 10 pf. I think it sounds pretty good and I don't want to desolder it again if I don't have to. I'm wondering if this is causing a rolloff into the audio band now.

 

[Bo Deadly]

Calculating the corner frequency of the compensation capacitor is non-trivial but using 20pF is almost definitely fine there. Especially with a gain of at least 10.

Note that there is a mistake in the schematic of the vol buffer stage where the 10p should be connected directly to pin 6. So it's missing a dot.

 

[abbey road d enfer]

Does anyone have a ballpark calculation for the rolloff caused by the 10 pf capacitor between pin 1 and 6 in the Ta-7136p gain stages.

Quite difficult to calculate, even if a detailed schematic was available. Good practice in this domain is making sure the value is high enough to result in stability even at the lowest gain, and low enough to ensure enough gain margin at the highest frequency and highest gain.

I wasn't paying attention and accidentally replaced one with a 22 pf c0g instead of 10 pf.

It probably would not result in HF roll-of, but may result in slew-rate limiting at high level.
The only way to know is measure and compare with a normal channel.

 

[rjb5191]

Thanks all. Can anybody help me understand how the 10 pf between pin 1 and 6 works vs. the typical compensation cap between pin 6 and the inverting input, the principal of which I understand at a basic level. I understand that this job is done by "internal compensation" now. Why would increasing the 10 pf limit the slew rate at high power levels?

Someone over at Diyaudio thought that the ta7136p is capable of excellent performance but the datasheet led many astray:

https://www.diyaudio.com/community/threads/why-is-the-ta7136-so-bad-its-the-datasheet.386249/

 

[abbey road d enfer]

Thanks all. Can anybody help me understand how the 10 pf between pin 1 and 6 works vs. the typical compensation cap between pin 6 and the inverting input, the principal of which I understand at a basic level.

The capacitor between output and inverting input is not the compensation capacitor per se.
Compensation is applied at the gain stage (Vas). Most often it applies Miller capacitance between collector and base.
In the TA7136, compensation is a combination of treble leak via the 50pF cap that shunts the collector of Q4 (half of the input LTP) and Miller between the output and the base of Q8.

Why would increasing the 10 pf limit the slew rate at high power levels?

Since there is current NFB from the output to the base of Q8, open loop gain decreases at HF, so more current is needed for driving the output stage, it reaches a point where it exceeds what the input stage can deliver.

Someone over at Diyaudio thought that the ta7136p is capable of excellent performance but the datasheet led many astray:

I believe it's a sequel of the trend in the 1990's to produce HiFi equipment with ultra violet frequency response and stellar rise-time, which resulted in not-so-stable designs.

 

[rjb5191]

Thanks, so is there any advantage in increasing the 10 pf to 22 pf? I sometimes pickup fax-like interference. Will this make the opamp more stable if it was marginally stable before? Reduce noise?

 

[abbey road d enfer]

Thanks, so is there any advantage in increasing the 10 pf to 22 pf? I sometimes pickup fax-like interference. Will this make the opamp more stable if it was marginally stable before? Reduce noise?

It may make it more stable, however it's hard to predict because the two compesation caps tend to fight each other somehow.
I would think that interference issues are more likely due to layout and insufficient input protection.
For example, pin 1 of input xlr's are connected to the audio ground, which is bad practice (however current at the time it was designed). Output xlr's are correct, though.

 

[rjb5191]

Thank you,I guess I just need to not be lazy and get the scope out and see what it's doing with a square wave. That makes sense with the input XLR issue on the schematic. I just looked though and pin 1 is actually bonded to the chassis with a short wire on all XLR connections. So, this is actually correct, right?

 

[JohnRoberts]

Perhaps measure frequency response and if you have equipment measure HF distortion... two-tone IMD might be more revealing of HF compromises.

JR

 

[rjb5191]

I'm working my way up to being able to do that kind of testing. Im always tickled when you guys throw around these more advanced (at least for me) techniques so nonchalantly .

 

[JohnRoberts]

Measuring frequency response is a pretty basic troubleshooting technique. Two-tone HF IMD is slightly more advanced but pretty revealing about slew limiting/over compensation.

Attempting to predict frequency response from compensation cap values is far more difficult IMO.

JR