I am using a regulated DC lab power supply for my LM-3875 gainclone.
Is it possible to just omit the electrolytic caps and make a more 'minimal' gainclone?
By the way, did anyone compare the quality of the DC coming out from a regulated power supply and the conventional design, on the oscilloscope or other way?
Thank you in advance for your kind help.
Best wishes,
Russell SIT
Is it possible to just omit the electrolytic caps and make a more 'minimal' gainclone?
By the way, did anyone compare the quality of the DC coming out from a regulated power supply and the conventional design, on the oscilloscope or other way?
Thank you in advance for your kind help.
Best wishes,
Russell SIT
Russell Sit said:
If you reduce the bypass capacitance too much you run a very real risk of oscillation. If you don't have an oscilloscope to measure the effects I wouldn't try it. Personally, I wouldn't try it anyway. I just re-read the section on supply bypassing on the LM-3875 data sheet, and maybe it's because I haven't had my coffee yet, but it wasn't particularly clear. They seem to recommend anywhere from 10//.1uF to 470uF LOCAL supply bypassing depending on how you read it. (This isn't the same as the bulk capacitance used at the bridge rectifier.)
I've looked at the output of regulated supplies, and the cleanliness varies widely amongst the brands. Some cheap ones are really good, some cheap ones are really bad, some expensive ones are mediocre and some are amazing. It all depends on the model.
Scott
Thank you for your reply, Scott.
The power supply is a Goodwill Instek GPC-1850D. I am running my gainclone at +/- 18v dc, which is probably the lowest voltage used among the other gainclones here.
Though manufacturer's published spec may not represent the actual performance, following is the from the spec sheet:
Constant Voltage Operation:
Line Regulation: <0.01% +3mV
Load Regulation: <0.01% +3mV (rating current £3A)
<0.02% +5mV (rating current £10A)
Ripple and Noise: <1mVrms 5Hz ~ 1MHz
Recovery Time: <100µs (50% load change, minimum load 0.5A)
I do not have a scope and so have no idea whether it is up to spec or not.
I raised this question because of the high power supply rejection ratio of the LM3875. If my power supply is good enough, probably I can eliminate the electrolytic caps, and the gainclone is then basically a chip with 2 resistors plus the dc power supply. Alternatively, if capacitance is still needed but of a much smaller value, then I will probably give those big polystryene capacitors a try.
Thank you very much for reading this thread.
Best wishes,
Russell SIT
The power supply is a Goodwill Instek GPC-1850D. I am running my gainclone at +/- 18v dc, which is probably the lowest voltage used among the other gainclones here.
Though manufacturer's published spec may not represent the actual performance, following is the from the spec sheet:
Constant Voltage Operation:
Line Regulation: <0.01% +3mV
Load Regulation: <0.01% +3mV (rating current £3A)
<0.02% +5mV (rating current £10A)
Ripple and Noise: <1mVrms 5Hz ~ 1MHz
Recovery Time: <100µs (50% load change, minimum load 0.5A)
I do not have a scope and so have no idea whether it is up to spec or not.
I raised this question because of the high power supply rejection ratio of the LM3875. If my power supply is good enough, probably I can eliminate the electrolytic caps, and the gainclone is then basically a chip with 2 resistors plus the dc power supply. Alternatively, if capacitance is still needed but of a much smaller value, then I will probably give those big polystryene capacitors a try.
Thank you very much for reading this thread.
Best wishes,
Russell SIT
The 3875 powered by regulated supply is certainly not a gainclone. Nor is it minimal. It may be better, it may be worse, but a gainclone it aint. Very few regulated supplies will do justice to a good GC. While there will no doubt be improvements in the bass and image stability, many other areas will suffer. And no, the oscilloscope won't tell you much.
cheers
peter
Hi,
Sorry, but I tend to disagree to a point.
It sounds like a sweeping generalistion even though I'm sure you didn't mean it that way.
As far as electrolytic caps in the PSU go, yes, I'd use them to the best of their benefit...their imperfections are such a great asset for filtering out unwanted interferences.
Horses for courses fellas..
Sorry, but I tend to disagree to a point.
It sounds like a sweeping generalistion even though I'm sure you didn't mean it that way.
As far as electrolytic caps in the PSU go, yes, I'd use them to the best of their benefit...their imperfections are such a great asset for filtering out unwanted interferences.
Horses for courses fellas..
hmmm...since russel hasint been back to acknowledge any replies I guess I wont give him any feedback LOL
But if anyone cares I would atleast use some 470 uf caps and bypassed with some 0.1 uf caps very close to the chip to smooth out any transients that are picked up between the power supply and amplifier. I am a big fan on tantulum caps near the supply pins as I have had great results with them.
DIRT®
But if anyone cares I would atleast use some 470 uf caps and bypassed with some 0.1 uf caps very close to the chip to smooth out any transients that are picked up between the power supply and amplifier. I am a big fan on tantulum caps near the supply pins as I have had great results with them.
DIRT®
I'm sure Russ will soon be back to check in on his question. Things here seem to stay dormant and then explode with activity when one is not looking. . .
As I understand the question, you are using a packaged regulated DC supply to power the op-amp (as by conventional means this isn't a GC). From my reading here, a GC would require minimal filtering, due to the chip's high rejection ratio. Proponents could say this makes the amp feel 'faster' whereas those that do a more robust filtering job (larger caps) seem to feel that the bass improves.
I would suspect that if it is a lab supply, it probably has a lot of built in filtering, so I would suspect that additional capacitance would be redundant.
I think there are a lot of ways to make an amp based on the
LM 3XXX or similar op-amps. I think the term GC invokes many emotions from various people, so you might receive many different responses.
I think, technically, to make what is considered a GC, you need small filter caps right at the chip, and that's it for filtering. To make an implementation of a chipamp which you can readily enjoy, a DC regulated supply could be a viable source, while it might not be considered optimum by many.
Sandy.
As I understand the question, you are using a packaged regulated DC supply to power the op-amp (as by conventional means this isn't a GC). From my reading here, a GC would require minimal filtering, due to the chip's high rejection ratio. Proponents could say this makes the amp feel 'faster' whereas those that do a more robust filtering job (larger caps) seem to feel that the bass improves.
I would suspect that if it is a lab supply, it probably has a lot of built in filtering, so I would suspect that additional capacitance would be redundant.
I think there are a lot of ways to make an amp based on the
LM 3XXX or similar op-amps. I think the term GC invokes many emotions from various people, so you might receive many different responses.
I think, technically, to make what is considered a GC, you need small filter caps right at the chip, and that's it for filtering. To make an implementation of a chipamp which you can readily enjoy, a DC regulated supply could be a viable source, while it might not be considered optimum by many.
Sandy.
Russell Sit said:
While the part does have very high PSRR, it is also a wide bandwidth, high gain, high power, high current device. All these things indicate instability when noise is about. Noise can creep in through a bad solder joint, wiring (even when it's done well), speaker cables, etc.
In electronics if a chip vendor recommends a part somewhere I tend to take that seriously, as 99.9999% of their customers are looking to save a 1/15 of a penny by pulling this part off here, and that part off there. If you have too many external components compared to your competitor, you will lose market share.
Since you don't have an o-scope, I would be more than hesitant to pull of those caps.
Scott
scott I agree with you to a certian extent but they are not planning on you using a regulated source but just a raw power source.
sandy I see your concerns and MOST dc power supplies encorparate fold back current limiting wich an amplifier will not like because of voltage drop on the rails....in fact they are perfect for testing because they will go to a dead short and not damage components being tested in most cases
DIRT®
sandy I see your concerns and MOST dc power supplies encorparate fold back current limiting wich an amplifier will not like because of voltage drop on the rails....in fact they are perfect for testing because they will go to a dead short and not damage components being tested in most cases
DIRT®
JOE DIRT® said:
Joe,
Maybe you realize this, but I'm reponding for Russel's benefit. A 20dB gain gainclone chip will have decent response past 200kHz. If you look in this area the PSRR of the chip isn't the 100dB or 120dB number the like to put in the data sheet table. It's around 30dB. At 10MHz and above, there is no rejection on the negative supply. None. Luckily at this point the gain of the amp will be less than 1.
Now, regulated supplies keep a nice regulated voltage independant of mains variation. They can also have handy current limiting features that make troublshooting much less destructive. They also have a wider bandwith of noise voltage since for all practical purposes they are a DC amplifier with a loop gain and loop bandwidth product of their own.
For anecdotal evidence, the worst hf instabillity I've ever seen in an audio amp (I've had a 1GHz bandwidth 10 watt RF amp go unstable and that was spectacular. I'm sure I was blocking radio stations for a very long way.) I've ever seen was with a regulated supply in a college lab. These guys didn't have a lick of bypass on an otherwise functional circuit. They added their caps (like the lab manual said too) and no more pffffzpt lost output devices. Now this circuit was a discrete amp with a design that had relatively little PSRR in it.
Scott
Sorry for the long delay
Hi!
First of all, I have to apologize for the long delay in my reply. I cannot agree with Sandy more: "Things here seem to stay dormant and then explode with activity when one is not looking. . . ". I thought it was already a dead thread when I saw the somewhat "philosophical" posting from Analog_sa. (no punch intended).
Thank you to all of you for your kind input. From what I have read in this thread, it seems that an important function of the caps is to filter out the high frequency noise. If I do not need a high capacitance to smooth out the DC supply, then can I use a much smaller value, e.g. 10uF, for the high frequecncy noise filtering?
By the way, Mr. Joe Dirt, could you please share with us your experience on the Tantalum caps in GC? How much capacitance have you used?
Once again, thank you all for your input and time in reading this thread.
Best wishes,
Russell SIT
Hi!
First of all, I have to apologize for the long delay in my reply. I cannot agree with Sandy more: "Things here seem to stay dormant and then explode with activity when one is not looking. . . ". I thought it was already a dead thread when I saw the somewhat "philosophical" posting from Analog_sa. (no punch intended).
Thank you to all of you for your kind input. From what I have read in this thread, it seems that an important function of the caps is to filter out the high frequency noise. If I do not need a high capacitance to smooth out the DC supply, then can I use a much smaller value, e.g. 10uF, for the high frequecncy noise filtering?
By the way, Mr. Joe Dirt, could you please share with us your experience on the Tantalum caps in GC? How much capacitance have you used?
Once again, thank you all for your input and time in reading this thread.
Best wishes,
Russell SIT
there have been alot of valid points been made here.
Russel I will explain how I have my GC setup for one channel......I have my ps located remotely from the amp with 2 15 000 uf Elna caps (one per rail) bypassed with .047 uf "orange drop" caps close to the amplfier pins I have 1 uf tantulum caps on each rail.
As it stands right now I am not hearing any audible noise with the current layout.....I am trying different arrangements on this one channel right now before I box it up and make it look pretty.
As far as my circuit goes I am using a 100k volume pot with a 1.5uf poly input cap and a 10 k input resistor and a 220k feedback....none of the parts are considered "audio grade".
I used point to point wiring and kept leads as short as possible
Cheers!!The DIRT®
Russel I will explain how I have my GC setup for one channel......I have my ps located remotely from the amp with 2 15 000 uf Elna caps (one per rail) bypassed with .047 uf "orange drop" caps close to the amplfier pins I have 1 uf tantulum caps on each rail.
As it stands right now I am not hearing any audible noise with the current layout.....I am trying different arrangements on this one channel right now before I box it up and make it look pretty.
As far as my circuit goes I am using a 100k volume pot with a 1.5uf poly input cap and a 10 k input resistor and a 220k feedback....none of the parts are considered "audio grade".
I used point to point wiring and kept leads as short as possible
Cheers!!The DIRT®
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.