Capacitors on THS4032
I am considering replacing on old RC4136 quad op-amp with the odd pin-outs with two THS4032 op-amps in a BrownDog adaptor. Celestian suggested using a 1uF Tantalum Cap from pins 8 to 4 and Radian sugessted a more optimized circuit using a 0.1 uF film and a 100-200 uF lytic from each power supply pin to ground keeping the leads as short as possible. If I go for the more simplifed approach from pins 8 to 4 do I need to apply caps to each IC or can i just place one cap from Vcc+ to Vcc- on the adaptor board to cover both IC's? My concern with adding the multiple caps as in Radian's approach is finding a ground point close to the IC socket. There doesn't appear to be a ground point on the adaptor board.
The application, for what it's worth, is to bring new life to a vintage Spectro Acoustic preamp. Unfortunately, I don't understand all I know about this topic, so your comments will be most appreciated.
I am considering replacing on old RC4136 quad op-amp with the odd pin-outs with two THS4032 op-amps in a BrownDog adaptor. Celestian suggested using a 1uF Tantalum Cap from pins 8 to 4 and Radian sugessted a more optimized circuit using a 0.1 uF film and a 100-200 uF lytic from each power supply pin to ground keeping the leads as short as possible. If I go for the more simplifed approach from pins 8 to 4 do I need to apply caps to each IC or can i just place one cap from Vcc+ to Vcc- on the adaptor board to cover both IC's? My concern with adding the multiple caps as in Radian's approach is finding a ground point close to the IC socket. There doesn't appear to be a ground point on the adaptor board.
The application, for what it's worth, is to bring new life to a vintage Spectro Acoustic preamp. Unfortunately, I don't understand all I know about this topic, so your comments will be most appreciated.
Hi Jmervin,
first look at the manufacturer's recomendations. That is your minimum requirement (for each opamp).
I would apply V+ to V- on each opamp.
I would also apply V+ to gnd and V- to gnd using about 100nF ceramic for each opamp.
I would apply V+ to gnd and V- to gnd using an electrolytic near the adaptor. Closeness is not quite so important at the lower frequencies that the electro is working at.
That's a total of 8 caps. You may find that all are not needed or do not offer any improvement.
first look at the manufacturer's recomendations. That is your minimum requirement (for each opamp).
I would apply V+ to V- on each opamp.
I would also apply V+ to gnd and V- to gnd using about 100nF ceramic for each opamp.
I would apply V+ to gnd and V- to gnd using an electrolytic near the adaptor. Closeness is not quite so important at the lower frequencies that the electro is working at.
That's a total of 8 caps. You may find that all are not needed or do not offer any improvement.
Hello Jmervin, two things:
The THS is not picky.
Ceramic capapacitors sound bad.
Ergo: to make the THS perform well just place a decent electrolytic cap from each power supply pin to ground, value 10uF is already OK but above 100uF there is definitely further improvement in bass behaviour.
For even better treble you can place a decent film cap (anything from 10nF to 1uF when there's space) parallel to each lytic. Avoid ceramics at all cost.
For the utmost finesse place a high value -like 220uF- lytic across + and -. Be aware that the voltage across these points is the sum total of two separate voltages, so at least take a 35V cap.
Bypass this with a good film cap see above further improves.
Ground can be found near the original IC. Short paths are overrated because we are not talking RF here, just audio and slightly above. Long paths in audio are considered above two inches -5cm- or so.
All improvements in capacitor quality translate linearly into better sound. But with the new opamp in place your sound will be better than before anyhow, so just go for what's physically achievable.
Cheers Marcel
The THS is not picky.
Ceramic capapacitors sound bad.
Ergo: to make the THS perform well just place a decent electrolytic cap from each power supply pin to ground, value 10uF is already OK but above 100uF there is definitely further improvement in bass behaviour.
For even better treble you can place a decent film cap (anything from 10nF to 1uF when there's space) parallel to each lytic. Avoid ceramics at all cost.
For the utmost finesse place a high value -like 220uF- lytic across + and -. Be aware that the voltage across these points is the sum total of two separate voltages, so at least take a 35V cap.
Bypass this with a good film cap see above further improves.
Ground can be found near the original IC. Short paths are overrated because we are not talking RF here, just audio and slightly above. Long paths in audio are considered above two inches -5cm- or so.
All improvements in capacitor quality translate linearly into better sound. But with the new opamp in place your sound will be better than before anyhow, so just go for what's physically achievable.
Cheers Marcel
Radian said:
as can be seen in my attachment
THS4032 op-amp is an almost identical toplogy circuit
to some Analog Devices AD80xx HiFi Operational Amplfiers.
Input, VAS + Diamond Buffer, it is all the same.
So I have no doubt this is an op-amp with great performance, even in Audio.
Regards, lineup
Attachments
Andrew
about what?
LineUp,
indeed, similar to AD8021 but not at all similar to AD8027, AD8031/32, AD8045 etc.
And between AD8021 and THS there are differences too: input LTP is cascoded in THS and plain collector resistors in AD.
Where the AD8022 has a cascode again funny enough.
So yes the basics are similar but the execution...not all the way-.
THS is still better for audio because of lower noise, especially in the audio region(!), more class A of the output stage , and higher output current so better drive and lower distortion at all loads especially feedback drive, cable drive, driving crazy loads like muting transistors and capacitive loads.
Maybe the AD8099 beats it by the way. Who tried?
about what?
LineUp,
indeed, similar to AD8021 but not at all similar to AD8027, AD8031/32, AD8045 etc.
And between AD8021 and THS there are differences too: input LTP is cascoded in THS and plain collector resistors in AD.
Where the AD8022 has a cascode again funny enough.
So yes the basics are similar but the execution...not all the way-.
THS is still better for audio because of lower noise, especially in the audio region(!), more class A of the output stage , and higher output current so better drive and lower distortion at all loads especially feedback drive, cable drive, driving crazy loads like muting transistors and capacitive loads.
Maybe the AD8099 beats it by the way. Who tried?
Hi Marconi,
Avoid Hi-K if possible.
Avoid modulated voltage.
Use to attenuate very fast transients.
Med-K are ideal for bypass at the main current consumers, particularly when the current is controlled in ClassAB fashion.
Low-K=NP0=C0G are usually too small here.
The RF filter at the input usually sees very low voltage so the capacitance variation, of an NP0, with signal voltage will also be low. The capacitance is of the order of 100pF to 680pF and this only cuts the frequencies above the audio band (680uS is -3db @ 234kHz and -0.1db @ about 25kHz, a +-10% variation in capacitance would be +-0.01db @ 25kHz). The slight variation in attenuation at these very high frequencies is inaudible even though the capacitance variation is at audio frequencies (IMD may be an issue but at an even lower level). The main advantage of ceramic here is the continued attenuation to frequencies that other caps cannot reach. If this bothers you, or you have proved you can hear the ceramic then use 90% as polypropylene and the remaining 10% as ceramic and locate the ceramic (and it's resistor) at the input terminal, again on short leads. A similar argument could be applied to feedback/forward caps, but unfortunately we are in the audio band so audibility may be an issue, continued attenuation to beyond the MHz range is vital.
when the wrong type are used or are used in an inappropriate location.
Avoid Hi-K if possible.
Avoid modulated voltage.
Use to attenuate very fast transients.
Med-K are ideal for bypass at the main current consumers, particularly when the current is controlled in ClassAB fashion.
Low-K=NP0=C0G are usually too small here.
The rapid changes in current inherent in ClassAB, inject very fast transients into the supply rail. These can be pulses or glitches and the sooner they are attenuated the better. Small ceramics mounted on the device would be ideal but not often achievable. Short paths are paramount.
The RF filter at the input usually sees very low voltage so the capacitance variation, of an NP0, with signal voltage will also be low. The capacitance is of the order of 100pF to 680pF and this only cuts the frequencies above the audio band (680uS is -3db @ 234kHz and -0.1db @ about 25kHz, a +-10% variation in capacitance would be +-0.01db @ 25kHz). The slight variation in attenuation at these very high frequencies is inaudible even though the capacitance variation is at audio frequencies (IMD may be an issue but at an even lower level). The main advantage of ceramic here is the continued attenuation to frequencies that other caps cannot reach. If this bothers you, or you have proved you can hear the ceramic then use 90% as polypropylene and the remaining 10% as ceramic and locate the ceramic (and it's resistor) at the input terminal, again on short leads. A similar argument could be applied to feedback/forward caps, but unfortunately we are in the audio band so audibility may be an issue, continued attenuation to beyond the MHz range is vital.
Re: Capacitors on THS4032
You should be careful with this replacement. THS4032 is a good opamp, however it has considerably higher bias current (3uA against 40nA typical, i.e. 75 times more), about 4 times more (for 4 amplifiers) current consumption and hence 4 times power dissipation and power supply load. THS4032 is also not unity gain stable and may oscillate if the circuit requires a unity gain stability. I would use OPA2134 in place of RC4136 as it would work without much circuit mods (different pinout!) (thought it would still ask for twice the power) or even TL074 - it needs about the same current and if the supply voltage is more than +/- 9V, would probably be much better than the original opamp. Otherwise you have to look carefully at the circuit diagram to see if you at all can replace RC4136 with 2xTHS4032. It could be a bad idea.
Cheers
x-pro
P.S. - here is the datasheet:
http://staff.bcc.edu/eet/4136-rca.pdf
jmervin said:
You should be careful with this replacement. THS4032 is a good opamp, however it has considerably higher bias current (3uA against 40nA typical, i.e. 75 times more), about 4 times more (for 4 amplifiers) current consumption and hence 4 times power dissipation and power supply load. THS4032 is also not unity gain stable and may oscillate if the circuit requires a unity gain stability. I would use OPA2134 in place of RC4136 as it would work without much circuit mods (different pinout!) (thought it would still ask for twice the power) or even TL074 - it needs about the same current and if the supply voltage is more than +/- 9V, would probably be much better than the original opamp. Otherwise you have to look carefully at the circuit diagram to see if you at all can replace RC4136 with 2xTHS4032. It could be a bad idea.
Cheers
x-pro
P.S. - here is the datasheet:
http://staff.bcc.edu/eet/4136-rca.pdf
Hi Andrew,
Most of what you say is true, especially scientifically true... not all is, though.
1: When classAB output stages are used, IE. cheap opamps, then ceramics get indeed rapid pulses to cope with. Note: these pulses are in swing with the audio signal. And ceramics will and do be modulated by this, and convert these currents into piezoelectric movement and finally into voltage. And this voltage is modulating your supply line. This you will and do hear at the output of the opamp.
au revoir good sound.
This is why we all should use only good opamps with class A output stages.
2: The modulation of the small LPF cap at inputs is proportional to the signal level. Signal level high or low, the result is the same.
Indeed NPO/COG is better than X7R in this respect but a good film cap, especially the non-inductive types, do a much better job here. Even in the MHz region, up to 100MHz even. And there are 1206 and 0805 film caps that go beyond that.
Tens or hundreds of MHz is no issue here, those signals are either amplified or attenuated. As long as they are not rectified, they cause no harm, well, generally.
Indeed WHEN they are rectified, then you're in trouble but that indicates that your active electronics are of disputable reputation, and/or your board layout has parasitics that cause feedforward or positive feedback in the hundreds of MHz region.
But it is presumed that this is not the case when you have substituted the opamps and what else have you there for the best you can buy.
So much for the theory.
Now the practice:
Eliminate all your ceramics from your signal chain and power supply lines. Have good films in their place.
Give it two weeks to settle. Get accustomed to the sound. Especially listen to sibilants, and the place in space they originate from. Should be the same spot as where you hear the mouth that sings or talks.
Then insert ONE ceramic, at whatever place, be it signal path or power supply pin of opamp.
Now locate the sibilants and note the difference. Let it play for 2 weeks and reassure yourself if the sibilants have moved from the mouth. If there is no difference, then try another place for the ceramic.
But I bet you'll hear it instantly...
Greetings
Marcel
Most of what you say is true, especially scientifically true... not all is, though.
1: When classAB output stages are used, IE. cheap opamps, then ceramics get indeed rapid pulses to cope with. Note: these pulses are in swing with the audio signal. And ceramics will and do be modulated by this, and convert these currents into piezoelectric movement and finally into voltage. And this voltage is modulating your supply line. This you will and do hear at the output of the opamp.
au revoir good sound.
This is why we all should use only good opamps with class A output stages.
2: The modulation of the small LPF cap at inputs is proportional to the signal level. Signal level high or low, the result is the same.
Indeed NPO/COG is better than X7R in this respect but a good film cap, especially the non-inductive types, do a much better job here. Even in the MHz region, up to 100MHz even. And there are 1206 and 0805 film caps that go beyond that.
Tens or hundreds of MHz is no issue here, those signals are either amplified or attenuated. As long as they are not rectified, they cause no harm, well, generally.
Indeed WHEN they are rectified, then you're in trouble but that indicates that your active electronics are of disputable reputation, and/or your board layout has parasitics that cause feedforward or positive feedback in the hundreds of MHz region.
But it is presumed that this is not the case when you have substituted the opamps and what else have you there for the best you can buy.
So much for the theory.
Now the practice:
Eliminate all your ceramics from your signal chain and power supply lines. Have good films in their place.
Give it two weeks to settle. Get accustomed to the sound. Especially listen to sibilants, and the place in space they originate from. Should be the same spot as where you hear the mouth that sings or talks.
Then insert ONE ceramic, at whatever place, be it signal path or power supply pin of opamp.
Now locate the sibilants and note the difference. Let it play for 2 weeks and reassure yourself if the sibilants have moved from the mouth. If there is no difference, then try another place for the ceramic.
But I bet you'll hear it instantly...
Greetings
Marcel
Valid point x-pro! Thanks for sharing this.
On the other hand, not many power supplies are getting upset by 40mA more current draw.
This changes when you are using this opamp in a 8- or 16-channel microphone mixer which I did. Had to build a bigger PSU...
But it definitely was worth it! What a difference.
Cheers
Marcel
On the other hand, not many power supplies are getting upset by 40mA more current draw.
This changes when you are using this opamp in a 8- or 16-channel microphone mixer which I did. Had to build a bigger PSU...
But it definitely was worth it! What a difference.
Cheers
Marcel
Thanks everyone for the input. I still am a bit confused on one point, however. I understand the use of the various caps from V+ and V- to ground as shown by TI on the datasheet for each op-amp. Since I will be mounting two of these IC's to the BrownDog adaptor board, this board has one pin each for V+ and V- with both IC's electrically tied to these points. Would not a single cap from each of the two supply pins on the adaptor to ground serve the same purpose as a cap from each IC's supply pins to ground? Electrically it's the same point. Am I missing something basic here? This would reduce the required number of caps by half and clean things up considerably from a real estate standpoint. I could see maybe a different value? This then begs the follow up question regarding the cap from V+ to V-. One for the adaptor board or one for each IC?
It sounds like maybe I could avoid a lot of this by using, say, the OPA2134, but I'm going for the maximum improvement in sound quality (subjective, I realize) and it sounds like the THS is pretty widely acclaimed as better sounding when applied correctly. As for the extra load on the power supply, I was planning on increasing the size of the PS filter caps and perhaps replacing the 1n4001 rectifier diodes with something faster and/or heftier. I'm not sure yet what the size of the transformer is. Haven't found many usable markings on it so far.
It sounds like maybe I could avoid a lot of this by using, say, the OPA2134, but I'm going for the maximum improvement in sound quality (subjective, I realize) and it sounds like the THS is pretty widely acclaimed as better sounding when applied correctly. As for the extra load on the power supply, I was planning on increasing the size of the PS filter caps and perhaps replacing the 1n4001 rectifier diodes with something faster and/or heftier. I'm not sure yet what the size of the transformer is. Haven't found many usable markings on it so far.
draw the current loop - if you drive a grounded load, or your output destabilzing parasitics are ground referenced then you want bypass caps to shorten any such loops at high frequencies
in most audio apps the bypass caps should to go to the circuit ground node
in most audio apps the bypass caps should to go to the circuit ground node
Post #47 by Radian
"...........Just finished 2 soundcards, M-audio 192 and my own Juli@. No change of AD/DA's (almost similar AKM's), but replaced clocks, opamps ......."
What opamp did you use to replace opamps on the Juli@ sound card ?
This is important to me as I have some blown ones on my card to replace . I couldn't find the original anywhere. I probably need just two or three of them.
Thanks.
Ashok.
"...........Just finished 2 soundcards, M-audio 192 and my own Juli@. No change of AD/DA's (almost similar AKM's), but replaced clocks, opamps ......."
What opamp did you use to replace opamps on the Juli@ sound card ?
This is important to me as I have some blown ones on my card to replace . I couldn't find the original anywhere. I probably need just two or three of them.
Thanks.
Ashok.
THS4031 AD8021
I didn't have time to look it up, which Analog Devices model was same as THS4031.
But thanks, it is AD8021
----------
As I see it, the main issue with using THS4031
is that for best Results In Audio
we need fairly LOW input resistances (100-1000 Ohm).
This is a feature THS4031 share with some other good Op-Amps.
AD797 for example.
... you maybe even should need to use an input buffering amplifier
OPA134, OPA2134 and other Op-Amps intended for Audio use
will of course work well with normal 10 - 100 kOhm input resistances.
Otherwise they wouldn't be 'Audio Amplifiers'.
lineup
marconi said:
I didn't have time to look it up, which Analog Devices model was same as THS4031.
But thanks, it is AD8021
----------
As I see it, the main issue with using THS4031
is that for best Results In Audio
we need fairly LOW input resistances (100-1000 Ohm).
This is a feature THS4031 share with some other good Op-Amps.
AD797 for example.
... you maybe even should need to use an input buffering amplifier
OPA134, OPA2134 and other Op-Amps intended for Audio use
will of course work well with normal 10 - 100 kOhm input resistances.
Otherwise they wouldn't be 'Audio Amplifiers'.
lineup
I just have some THS4032's to try.
What ps should i build, or wich regulator is best then, instead of the mostly used LM78** supply to get best performance from THS4032?
What ps should i build, or wich regulator is best then, instead of the mostly used LM78** supply to get best performance from THS4032?
OPA2134 versus LM4562
The National LM4562 SMOKES the OPA2134 in specs for THD,IMD,CMRR, and Noise Figure. The 2134 is an FET design, however, for those who prefer that, and it does have a greater voltage swing possible--- it can be operated on =/-18v; the 4562 is speced only to =/-17v. BUT, the LM4562 is also much less expensive.
The National LM4562 SMOKES the OPA2134 in specs for THD,IMD,CMRR, and Noise Figure. The 2134 is an FET design, however, for those who prefer that, and it does have a greater voltage swing possible--- it can be operated on =/-18v; the 4562 is speced only to =/-17v. BUT, the LM4562 is also much less expensive.
Hi Marcel!
I´ve just replaced four OpAmp for an Harman Kardon HD970, two OP275 for I/V and two OPA2134 for buffer/signal out.
I soldered a 0,1uF for each THS4032 between pin 4 and 8 (power pins), but after reading your post´s, I realize there are a 100nF (ceramic cap, I think) to ground only for the two I/V´s, no cap for the buffer´s.
So, do you think I should replace the two 100nF with 0,1uF plastic film bypassed with a 220uF electrolyte for all of the THS4032 (gonna need eight set´s, as there are four IC´s)?
Or may I leave it as it is, with just a 0,1uF film cap for each?
If replacing, should the cap´s negative - be faced to ground and positive + to IC?
Best regards: Mikael Markstrom
Sorry, I was looking at the schematic and each OpAmp were printed two times as they´re dual. But now I realize the I/V has a 100nF for positive and a 100nF for negative voltage. And the buffer/output having 100nF and also a 47uF electrolyte for each voltage pin.
All 0,1uF are ceramic, and there are a total of 55 pcs ceramic for the design - all between 100pF and 0,1uF. Are the cap´s for voltage to ground more picky with ceramic´s or should every ceramic cap be replaced with bipolar plastic one´s?
If I let everything be as it is now, and with the 0,1uF between pin 4 and eight for each OpAmp, will it be bad?
- Status
- Not open for further replies.