I have been using passive pre-amp in my listening environment. Since the volume pot is 50 kA, it can cause undesirable high-freq roll off for my 498xx based power amp (input 33k//470p). I finally made a unity gain pre-amp (buffer). It is just a quick project where an old spare PCB is used to populate the opamp circuit. The board was designed for signal conditioning with Linear active filter chips. The 50 kA pot is followed by two OPA134 as unit-gain buffers. Additional output RCA sockets are added for driving bi-amp. Regulated +/- 15 V supply is from a separate supply chassis. LM317/337 and some caps form the simple supply. Everything just work great! No noise (hiss and hum) can be heard from the speaker with the pre-amp input grounded.
I will modify it slightly for some gain no greater than 10. I think this is very sufficient for today digital sources.
I will modify it slightly for some gain no greater than 10. I think this is very sufficient for today digital sources.
Attachments
First of all I would like to congratulate you on a job well done. I especially like the way you isolated the power supply in its own chassis. If you ever get tired of an opamp based unity gain buffer, please look into Salas direct coupled Class A buffer with shunt regulation right here. It is inspired by Nelson Pass' B1 and therefore only uses Jfets. I am building a differential/balanced version of that design.
Kind regards,
Anand.
Thank you Anand!
My buffer pre-amp shown in thread #1 is borrowed to a friend for listening evaluation. I use the TI opamp universal test board (fixture) to build another pre-amp. LME49720 is soldered to the board as the first test unit. The chip is still configured as a voltage-follower. A cancellation resistor is inserted between output and -IN. A 50 kA volume pot is followed by the opamp. Hence, the source resistance of the opamp is not a fixed value. A fixed cancellation resistor won't offer maximum cancellation for different pot positions.
Let see the THD vs freq for cancellation resistor equal to 1 k. The output level was kept at 3 Vrms in order to maintain signal to noise ratio. Output load is the AP input 100 k resistance. The curve for Rs = 0 is essentially the AP's THD vs freq.
Other plot is for cancellation resistor equal to 16 k. We can see that this is over kill! The distortion is worst for Rs = 0. Maximum (lowest distortion) cancellation occurs at Rs = 16.8 k. Well, Rs is nearly identical to the cancellation resistance.
Based on the plots, a 1 k cancellation resistor offers better performance. Taking listening level into account, 16 k is probably the choice. I will try 10 k next.
My buffer pre-amp shown in thread #1 is borrowed to a friend for listening evaluation. I use the TI opamp universal test board (fixture) to build another pre-amp. LME49720 is soldered to the board as the first test unit. The chip is still configured as a voltage-follower. A cancellation resistor is inserted between output and -IN. A 50 kA volume pot is followed by the opamp. Hence, the source resistance of the opamp is not a fixed value. A fixed cancellation resistor won't offer maximum cancellation for different pot positions.
Let see the THD vs freq for cancellation resistor equal to 1 k. The output level was kept at 3 Vrms in order to maintain signal to noise ratio. Output load is the AP input 100 k resistance. The curve for Rs = 0 is essentially the AP's THD vs freq.
Other plot is for cancellation resistor equal to 16 k. We can see that this is over kill! The distortion is worst for Rs = 0. Maximum (lowest distortion) cancellation occurs at Rs = 16.8 k. Well, Rs is nearly identical to the cancellation resistance.
Based on the plots, a 1 k cancellation resistor offers better performance. Taking listening level into account, 16 k is probably the choice. I will try 10 k next.
Attachments
Last edited:
The plot is for a cancellation resistor equal to 12 k. It sees this is a better value compared to the previous 1 k and 16 k. We have maximum (complete?) cancellation for the first three Rs values.
The test fixture is shown as well.
The test fixture is shown as well.
Attachments
Last edited:
The LME series is superb in supressing common mode distortion. I asume your OPA134 version whould show much bigger differences.
Yes, you are right that the LME outperforms OPA134 in this regard. I tested the OPA2134 for CM distortion. The cancellation resistor is still effective. However, the chip does not like high valued resistor where it oscillates. A 10 pF cap is parallel with the cancellation resistor to stop osc.
The plots for 1 k, 6 k and 12 k cancellation resistor are shown below.
Attachments
Very interesting measurements. Proves the concept and gives results with higher source impedance. Did you compare the sound with and without the resistor ?
Hi Joachim,
I am going to do listening test. I will report what I can hear (w and w/o resistor and chips).
I am going to do listening test. I will report what I can hear (w and w/o resistor and chips).
Great ! I think this is the first time somebody does that in public. I always try to avoid common mode distortion in my designs but actually i measure and did not do listening tests other then checking the finished circuit.
opamp follower
is the cancellation resistor a resistance in the normally seen shorting link of the opamp as (non-inverting) unity gain follower?
Normally we see Output connected to -IN as a direct short.
This has come up in a couple of other threads recently.
is the cancellation resistor a resistance in the normally seen shorting link of the opamp as (non-inverting) unity gain follower?
Normally we see Output connected to -IN as a direct short.
This has come up in a couple of other threads recently.
Andrew
This was done by Silicon Chip magazine in their Studio Series Preamplifier of October 2005 with an OPA2134.(P.29)
SandyK
Do you ware of the Samuel Groner opamp distortion report? http://www.sg-acoustics.ch/analogue_audio/ic_opamps/pdf/opamp_distortion.pdf
Yes. I know Groners work and of cause i know what Dougas Self has to say about it. Sofar i have only seen and done measurement and a listening test whould be really interesting.
Silicon Chip mentioned in their article to use a resistor that's the same value as the wiper to ground resistance on the pot at your normal listening level. On my buffers I just tend to use 10% of the pot value.
To necrobump this a bit :
How is the distortion affected by signal level, i e CM voltage, fex if you have a volume pot before the buffer? Here and in Selfs book the testing is done at quite high voltage.
I rarely feed 3Vrms into my power amps, as the distortion from the speakers probably outshine the buffer distortion by a couple of magnitudes then 🙂
How is the distortion affected by signal level, i e CM voltage, fex if you have a volume pot before the buffer? Here and in Selfs book the testing is done at quite high voltage.
I rarely feed 3Vrms into my power amps, as the distortion from the speakers probably outshine the buffer distortion by a couple of magnitudes then 🙂
- Status
- Not open for further replies.