RX5 said:hmmm...
lucpres,
link dont work....
http://myweb.tiscali.co.uk/nuukspot/decdun/gainclone2.html
The board added a quota inside the link.
Hi Lucpres,
been reading almost all relevant issues.... so this 990 is sort of like an LM386 eh.. It CAN directly drive a speaker (or low ohms) .... depending on output transistors....
NE5532 has also low impedance output.... hmmmmmm thats good, i think... compatible??
but I have to make 2 of this 990 to come up with a dual op-amp
in a week or two.. maybe, ill make one up ....
Cheers,
Raff
been reading almost all relevant issues.... so this 990 is sort of like an LM386 eh.. It CAN directly drive a speaker (or low ohms) .... depending on output transistors....
NE5532 has also low impedance output.... hmmmmmm thats good, i think... compatible??
but I have to make 2 of this 990 to come up with a dual op-amp
in a week or two.. maybe, ill make one up ....
Cheers,
Raff
rotation said:
Where I can find the specs of the UcD modules without the op-amp?
Is this op-amp inside any feedback loop?
Is it removable?
I have already what I deem a nice single ended line stage that I would like to use in order to drive these modules without the input op-amp for better sound.
It has voltage gain=6 and can drive comfortably impedances as low as 5kohm.
Is this feasible?
Is a balanced line stage mandatory ?
Any info would be extremely appreciated here.
Thank you very much and kind regards,
beppe
lucpes said:
RX5 said:@rotation,
since you told me you already have done the 990, have you tried swapping it in place with a dual op-amp(TL072,NE5532)???
I plan to make 2 DOA and swap with an NE5532, my main goal
-Raff
Lucpes: That is good idea, but do you think it's worth investing time in LM based amp? To my ears none of them sound good, at least if compared to discrete designs.
RX5: I tried 990 in API 3124 preamplifier. I think i could use OPA, NE with socket instate. What i know is that 990 sounds better than other two.
Hi Beppe61,
There is an application note on the Hypex site that you need to read here:
http://www.hypex.nl/docs/gainappnote.pdf
This shows that the UCD module is made up of two cascading sections. The first is a balanced-in / balanced out stage with a gain of about 4 or 5 made up of two op-amps. The second stage is the PWM comparator and output section. The impedance values you asked for are given. It also has a gain of 4.5. Each or the two sections has its own feedback structure. There is no overall feedback. You could drive the comparatot inputs directly but your gain would only be 12 to 14 dB. Not very much - but it might be enough for your system.
There is a Philips app note that contains a schematic which looks like it would be close to the UCD schematic - at least for your purposes. It is here:
http://www.nxp.com/acrobat_download/usermanuals/UM10155_1.pdf
A good source on discrete op-amp design from Nelson Pass is here:
http://www.passdiy.com/pdf/diyopamp.pdf
I am building a pair of UCD700AD monoblocks. I will run them with the 8620's to start with and I am then planning on replacing the front end with a discrete design. It will take several more months to complete the basic amps and I plan on posting the results on this site. It will take several more months beyond that to complete the discrete version and evaluate it.
I would like to encourage you in your efforts. There are other people (like me) that have the same goals you do. Please post the your results of your work as your project progresses.
Regards,
Graeme
There is an application note on the Hypex site that you need to read here:
http://www.hypex.nl/docs/gainappnote.pdf
This shows that the UCD module is made up of two cascading sections. The first is a balanced-in / balanced out stage with a gain of about 4 or 5 made up of two op-amps. The second stage is the PWM comparator and output section. The impedance values you asked for are given. It also has a gain of 4.5. Each or the two sections has its own feedback structure. There is no overall feedback. You could drive the comparatot inputs directly but your gain would only be 12 to 14 dB. Not very much - but it might be enough for your system.
There is a Philips app note that contains a schematic which looks like it would be close to the UCD schematic - at least for your purposes. It is here:
http://www.nxp.com/acrobat_download/usermanuals/UM10155_1.pdf
A good source on discrete op-amp design from Nelson Pass is here:
http://www.passdiy.com/pdf/diyopamp.pdf
I am building a pair of UCD700AD monoblocks. I will run them with the 8620's to start with and I am then planning on replacing the front end with a discrete design. It will take several more months to complete the basic amps and I plan on posting the results on this site. It will take several more months beyond that to complete the discrete version and evaluate it.
I would like to encourage you in your efforts. There are other people (like me) that have the same goals you do. Please post the your results of your work as your project progresses.
Regards,
Graeme
rotation,
I know its cheap, 10$ a piece..actually 2 for replacing NE5532... but My main goal is TO EXPERIMENT with locally available components.... just like my diy UCD...it might not have the actual parts used in the commercial units, but IT PERFORMS.. sont worry, i WILL post results... but that would have to be a few weeks from now.. need to look for parts that are available here... I really am impatient with mail orders.. coupled to that, I dont have much resource
I know its cheap, 10$ a piece..actually 2 for replacing NE5532... but My main goal is TO EXPERIMENT with locally available components.... just like my diy UCD...it might not have the actual parts used in the commercial units, but IT PERFORMS.. sont worry, i WILL post results... but that would have to be a few weeks from now.. need to look for parts that are available here... I really am impatient with mail orders.. coupled to that, I dont have much resource
I have always questioned why they went with the buffer design.
their topology, from what I read, is a differential design so you need to convert a SE input to differential output. That's understandable.
However, for each side of the differential amp, why did they go with an inverting topology? what will happen if you go with a non-inverting design? would that have worked?
why did they go with a differential design? the philips application note suggests that a SE design would work. does hypex want to get to a certain power rating with the design? why wouldn't they just increase the rail voltage to do that?
their topology, from what I read, is a differential design so you need to convert a SE input to differential output. That's understandable.
However, for each side of the differential amp, why did they go with an inverting topology? what will happen if you go with a non-inverting design? would that have worked?
why did they go with a differential design? the philips application note suggests that a SE design would work. does hypex want to get to a certain power rating with the design? why wouldn't they just increase the rail voltage to do that?
to add to it: the pwm comprator outputs to complimentary signals. Wouldn't it be easier to just use the signal to run two half bridges, in a complimentary fashion to double the output? for example, use one pwm out to drive the high side of one half bridge and the low side of another half bridge, and the other pwm output to drive the low-side of the 1st half bridge and the high side of the 2nd bridge?
wiring the feedback would be a challenge, though.
wiring the feedback would be a challenge, though.
The op-amp buffers are set up with the comparator/outputs as a classic instrumentation amp topology. This is a well established approach. It's cheap and simple and gives good results.
You can achieve better results with discretes. Hypex even says so themselves (somewhere).
A discrete solution needs to provide the same things the dual op-amps do. There are a number of solutions. The most foolproof is two homebrew op-amps built with discrete parts.
Graeme
You can achieve better results with discretes. Hypex even says so themselves (somewhere).
A discrete solution needs to provide the same things the dual op-amps do. There are a number of solutions. The most foolproof is two homebrew op-amps built with discrete parts.
Graeme
1) I posted the link just for the input buffers schemtics availbable there - driving the UcD's modulator input instead of the LM3886.
2) To directly drive the modulator just use the coupling capacitor holes.
3) NE5532 sounds like crap IMHO, most people agree that if AD8620 sounds too bright the problem is elsewhere in your system.
2) To directly drive the modulator just use the coupling capacitor holes.
3) NE5532 sounds like crap IMHO, most people agree that if AD8620 sounds too bright the problem is elsewhere in your system.
lucpes said:
lol @ #3..
regarding the changing of op-amps, I changed the JRC4558 with NE552 in my CD player... yup... I dont know If I could label it as TOO BRIGHT, but yeah.. a slight increase of treble though....as compared to the 4558....
I think it would be much brighter if I could change to AD8620?
fokker said:
Guess Hypex needs the fully differential design to get the upper mosfet to switch
fokker said:
You really need to look at it more until these answers come to you, and you realize how very elegant it really is, and I don't mean the Philips app note either. Consider that they don't care to give you all the answers either.
gl said:
Bruno said "his" discrete front ends sound better. What's most foolproof is leaving the stock op amp in, unless maybe you think you can outdo the performance of an AD8620 ...
Personally I would not be too quick to think I can simply download whatever schematic from the web and slap it in there, and honestly expect some improvement. No way. It'd have to be taken a little more seriously.
I can wholeheartedly agree with that, if the performance of an AD8620 can be matched with a few discrete devices thrown together on a piece of breadboard, then why isn't Analog Devices selling that instead?
I've seen people try to use a tube based preamp to drive the modulator directly at some point and that was laughable at best. The only implementation I've actually seen to work is Bruno's own discrete solution. He himself acknowledges that even that might not be better perse than a AD8620 if implemented correctly (CRDs, seperate opamp supply, no coupling caps, 1K8/3K3 mini-melf in the feedback).
In order to improve upon the basic circuit of the UcD you'll need to outsmart Bruno, and trust me (coming from someone that knows Bruno personally) that's going to take a whole lot more than some experience with modding equipment or a high-school diploma.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
gl said:
Dear Mr Graeme, thank you very much for the kind and helpful advice.
The first is a balanced-in / balanced out stage with a gain of about 4 or 5 made up of two op-amps.
This is actually the stage I want to replace with my line stage (voltage gain=6; very low output impedance).
The overall voltage gain could be more than enough.
My room is limited and my speaker are 4ohm, 87dB/1W.
Nice to know that I have just to desolder two coupling caps and solder the new wires.
I think I can do this with a little help from my friend (he has a very nice desoldering machine at hand).
Thank you so much again.
Kind regards from Italy,
beppe
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.