hi, i know these questions must be getting really old, but please bear with me, i'm a little new to all of this...
some questions:
1.i checked dxvideo's pcb layouts for the BPA - 200 and he hadn't used the servo circuits.really, what goes wrong if you don't include them???(makes the Pcb layout sooo much easier.
)
2.the AN - 1192 says use a 60vrms transformer.that means a 30 - 0 - 30 right??...so that means the final dc o/p should be around 27v. but how do they show 42v at the o/p, and that too unregulated...(as an off question, isin't unregulated inherently unstable??)
and finally, going out on a limb here, but what are the chanches any of you have some pcb layouts of the circuit? i'm finding the going pretty hard.using trax maker 2000.
thanks a lot.
some questions:
1.i checked dxvideo's pcb layouts for the BPA - 200 and he hadn't used the servo circuits.really, what goes wrong if you don't include them???(makes the Pcb layout sooo much easier.
)2.the AN - 1192 says use a 60vrms transformer.that means a 30 - 0 - 30 right??...so that means the final dc o/p should be around 27v. but how do they show 42v at the o/p, and that too unregulated...(as an off question, isin't unregulated inherently unstable??)
and finally, going out on a limb here, but what are the chanches any of you have some pcb layouts of the circuit? i'm finding the going pretty hard.using trax maker 2000.
thanks a lot.
1- I would forget about the servo's, they are not required and the stability they add is probably traded off by degradation in the sound. Just make sure your gain setting parts are well matched, and have some output resistors and i think you will be fine.
2- 30ACV x 1.4 is 42DCV. I would say this is pushing it a bit to high for power. You have no margin for error with main fluctuations, and your chips will be clipping/overheating straight away too. A 25V transformer would seem to be a much better choice, it still can’t quite handle a 4ohm load (requires .90c/w heat sink) but it’s better then the 30V. of course if you only plan on running this into an 8 ohm load, then you may be ok with something a bit higher.
As for PCB layout, PCB’s can be handy, but for something simple like this. I don't think you can beat nice point to point setup or just use some vario board.
2- 30ACV x 1.4 is 42DCV. I would say this is pushing it a bit to high for power. You have no margin for error with main fluctuations, and your chips will be clipping/overheating straight away too. A 25V transformer would seem to be a much better choice, it still can’t quite handle a 4ohm load (requires .90c/w heat sink) but it’s better then the 30V. of course if you only plan on running this into an 8 ohm load, then you may be ok with something a bit higher.
As for PCB layout, PCB’s can be handy, but for something simple like this. I don't think you can beat nice point to point setup or just use some vario board.
I don't want to be rude, but Officeboy are not correkt in most of his statements.
Have a look at
http://www.diyaudio.com/forums/showthread.php?s=&threadid=30570
before you start paralling.
Have fun
Thomas
Have a look at
http://www.diyaudio.com/forums/showthread.php?s=&threadid=30570
before you start paralling.
Have fun
Thomas
tlmadsen said:
Sorry to disagree, but Officeboy is correct in most of his claims, as I see it.
The way National suggests to go about the chip paralleling is interesting, but servos seem to have an effect on audio quality.
If the parts are matched, to within 0.1% on the gain resistors, and use slightly larger 5% resistors than those suggested by National for the output, things should be fine. Rowland used 0.200 ohm resistors on his first amp with the 3886 chips, and 0.500 ohms in the next.
The voltage suggested by Groyne (+/- 42VDC) is certainly high and can compromise the chip response. Paralleling chips is good for lower impedance speakers, but they may demand too much at that voltage going into protection or heat too much.
The only think I do not agree, for single or multiple chips, is doing p2p wiring. It is messy, can be prone to accidents and shouldn't improve on a properly made pcb. You can do short paths with a pcb too.
Carlos
thanks carlmart,
however, won't using higher valued resistors cause a much higher drop at the same power o/p.
if i wanted to reduce power dissipation across the resistors, i'd then have to reduce my o/p power, would'nt i?
and i think i'm gonna go the pcb way anyway.seems nicer, and troubleshooting would be much easier, i think( though ideally i should'nt need it. )
however, won't using higher valued resistors cause a much higher drop at the same power o/p.
if i wanted to reduce power dissipation across the resistors, i'd then have to reduce my o/p power, would'nt i?
and i think i'm gonna go the pcb way anyway.seems nicer, and troubleshooting would be much easier, i think( though ideally i should'nt need it.
)
Hmmmmmmmmmmmmmm
I think that we all are somehow on the same pace here, but I would like to make a few points clear.
When Officeboy say that the DC-servos are not needed he is basically right, BUT you have to deal with the DC-offset when ýou are paralling chips. This can be done by DC-servos, capacitors, multiturn-offset adjustment, bigger output resistors and so on.
However:
"Just make sure your gain setting parts are well matched, and have some output resistors and i think you will be fine"
will NOT do the job. First of all the "matching" of the gain setting parts have nothing to do with the DC-offset and making then "match" by 0.000001 % will NOT remove the DC-offset.
It is correct as carlmart states that making the output-resitors "big" can to some extent solve the problem of DC flowing between the chips. To make this a "complete" cure you have to make the output-resistor bigger that 0.5 ohm witch I think the the upper limit, otherwise you start to loose to much on efficiency and so on.
I don't what to start the old discussion about DC-servos or not, but I think officeboy is a little bit to fast stating that they degrading the sound without ever have tried them. There are actually difference between how DC-servos are implementet and the one used in BPA-200 are not bad.
The +/- 42 V DC is correctly on the high side, but unlees you have a BIIIIIIIIIG transformer (+500 VA) the voltage will drop when you start pushing current through your BPA-200. (try to calculate the current you need at max. outpout) With no signal thay can actyally take +/- 47 V DC.................and yes it is true, I have tried is.
If you have the need for the high power, use 8 ohm speakers and have plenty of cooling you will be fine with your 2 x 30 V AC transformer.
I think p2p wiring of a BPA-200 is possible, but it will fore sure not be better than a PCB and the changes for some going wrong are quit high.
Have fun
Thomas
I think that we all are somehow on the same pace here, but I would like to make a few points clear.
When Officeboy say that the DC-servos are not needed he is basically right, BUT you have to deal with the DC-offset when ýou are paralling chips. This can be done by DC-servos, capacitors, multiturn-offset adjustment, bigger output resistors and so on.
However:
"Just make sure your gain setting parts are well matched, and have some output resistors and i think you will be fine"
will NOT do the job. First of all the "matching" of the gain setting parts have nothing to do with the DC-offset and making then "match" by 0.000001 % will NOT remove the DC-offset.
It is correct as carlmart states that making the output-resitors "big" can to some extent solve the problem of DC flowing between the chips. To make this a "complete" cure you have to make the output-resistor bigger that 0.5 ohm witch I think the the upper limit, otherwise you start to loose to much on efficiency and so on.
I don't what to start the old discussion about DC-servos or not, but I think officeboy is a little bit to fast stating that they degrading the sound without ever have tried them. There are actually difference between how DC-servos are implementet and the one used in BPA-200 are not bad.
The +/- 42 V DC is correctly on the high side, but unlees you have a BIIIIIIIIIG transformer (+500 VA) the voltage will drop when you start pushing current through your BPA-200. (try to calculate the current you need at max. outpout) With no signal thay can actyally take +/- 47 V DC.................and yes it is true, I have tried is.
If you have the need for the high power, use 8 ohm speakers and have plenty of cooling you will be fine with your 2 x 30 V AC transformer.
I think p2p wiring of a BPA-200 is possible, but it will fore sure not be better than a PCB and the changes for some going wrong are quit high.
Have fun
Thomas
If you have them you can remove the offset and also the temperature drift plus you can also have smaller output resistors (creating a higher damping factor) plus a bass response to 1 Hz or below.groyne said:1.i checked dxvideo's pcb layouts for the BPA - 200 and he hadn't used the servo circuits.really, what goes wrong if you don't include them???(makes the Pcb layout sooo much easier.
If you choose good opamps for the servo I'd say that they won't degrade the sound quality very much.
"sound" of an integrator
Ok, I'm a new guy here, but I've been in audio for 30+ years.
Can some one explain to me how a lowpass circuit that starts to roll off at around 6 Hz could affect the sound of an amplifier when used for negative feedback? Do not the blocking caps do the same thing, but with out the advantage of the servo effect?
Ok, I'm a new guy here, but I've been in audio for 30+ years.
Can some one explain to me how a lowpass circuit that starts to roll off at around 6 Hz could affect the sound of an amplifier when used for negative feedback? Do not the blocking caps do the same thing, but with out the advantage of the servo effect?
The purpose of the DC servo is to remove the output offset voltage which is a problem if you want to connect amps in parallel AND keep the series resistors low AND keep low power consumption (to no use).
The DC servo is within a feedabck loop. Signals from the output is feed to the input via an opamp which of course must be good up to 1 kHz approx. If you do the deisng job correctly the impact of the DC servo is minor when it comes to sound quality. I for instance use a 1/2 of an AD8620 is servo, quite good. Using a LM324 is not good!
Example here
The DC servo is within a feedabck loop. Signals from the output is feed to the input via an opamp which of course must be good up to 1 kHz approx. If you do the deisng job correctly the impact of the DC servo is minor when it comes to sound quality. I for instance use a 1/2 of an AD8620 is servo, quite good. Using a LM324 is not good!
Example here
Servo me up
Right! Thanks for the reply.. sounds like you see this the way I do, I don't get how the servo could possibly degrade the sound.. I guess I just wanted an explanation of why some one would think it could. I really don't wan't to be controversial here, but I sure am seeing more and more commentary on how things "sound", right down to the kind of solder used.
I'm not blessed with "golden ears" so I am happy with clean, flat, dynamic circuitry, and sinking the lions share of my cash on great transducers (speakers, mics and disc players).
I really don't care if someone wants to spend $1000 on a cap or some speaker cables. It's their money after all. Sometimes, however, weeding through all the rhetoric can get just a little taxing!
Sorry about the venting...
Now I can go with servos and not look back!
Cheers
Right! Thanks for the reply.. sounds like you see this the way I do, I don't get how the servo could possibly degrade the sound.. I guess I just wanted an explanation of why some one would think it could. I really don't wan't to be controversial here, but I sure am seeing more and more commentary on how things "sound", right down to the kind of solder used.
I'm not blessed with "golden ears" so I am happy with clean, flat, dynamic circuitry, and sinking the lions share of my cash on great transducers (speakers, mics and disc players).
I really don't care if someone wants to spend $1000 on a cap or some speaker cables. It's their money after all. Sometimes, however, weeding through all the rhetoric can get just a little taxing!
Sorry about the venting...
Now I can go with servos and not look back!
Cheers
BPA 200
Servo's even well designed ones do impact the amplifiers sound quality ,IMHO and you have to wounder why people like Jeff Rowland didnt use them?? Have you ever thought why you dont see manufactures just making tons of high power BPA chipamps? they are easy to implement low parts count ect..... because with out servos they are not reliable and with servo's the design is still very important they increase the parts count and complexity makeing them unviable.
Servo's even well designed ones do impact the amplifiers sound quality ,IMHO and you have to wounder why people like Jeff Rowland didnt use them?? Have you ever thought why you dont see manufactures just making tons of high power BPA chipamps? they are easy to implement low parts count ect..... because with out servos they are not reliable and with servo's the design is still very important they increase the parts count and complexity makeing them unviable.
Tiltedhalo, do you have persnonal experience of the statement of yours or is it something you have heard?
The reason why we don't see lot's of LM3886 paralleled is the cost I think. It's better to have a class D module. Jeff Rowland for instance uses class D if I'm correct.
DC servos are reliabe if you design them right. Can you give me an example where they don't are reliable?
The reason why we don't see lot's of LM3886 paralleled is the cost I think. It's better to have a class D module. Jeff Rowland for instance uses class D if I'm correct.
DC servos are reliabe if you design them right. Can you give me an example where they don't are reliable?
servo
peranders, you misunderstood me I meant servo's will increase the reliability, but increase design complexity and cost making them costly for mainstream production.
I have experimented with servo's and have had some very good results, but prefer non servo design. One of my early stabs at a BPA used my self equalizing feed back followed by a global differential DC high speed comparator servo fed back to the cross coupled lines of the bridged driver, worked very well but lost some detail and "air" IMHO.
The problem with non servo'ed designs are stability no matter how close your parts are matched minute differences will become apparent as differences in current sharing resulting in waste even if as little as 5W or as great as 50w, and sonic pit falls.
By the way Roland used LM3886's in a inverting mode to set idle DC offset.
peranders, you misunderstood me I meant servo's will increase the reliability, but increase design complexity and cost making them costly for mainstream production.
I have experimented with servo's and have had some very good results, but prefer non servo design. One of my early stabs at a BPA used my self equalizing feed back followed by a global differential DC high speed comparator servo fed back to the cross coupled lines of the bridged driver, worked very well but lost some detail and "air" IMHO.
The problem with non servo'ed designs are stability no matter how close your parts are matched minute differences will become apparent as differences in current sharing resulting in waste even if as little as 5W or as great as 50w, and sonic pit falls.
By the way Roland used LM3886's in a inverting mode to set idle DC offset.
Just so we're on the same page, I was reffering to the servos in AN1192...
So even though there is nothing in the audio range getting back to the input through the servo, it can still effect sound quality?
I was hoping for some one to give me an actual explanation of how this can happen.
BTW I wont mention the manufacturer (don't want to be fodder for their army of lawers), I just had to repair a product that is mass produced, and contains a paralleled 3886 pair. No servos.
Standard DC blocking caps in the feedback...
So even though there is nothing in the audio range getting back to the input through the servo, it can still effect sound quality?
I was hoping for some one to give me an actual explanation of how this can happen.
BTW I wont mention the manufacturer (don't want to be fodder for their army of lawers), I just had to repair a product that is mass produced, and contains a paralleled 3886 pair. No servos.
Standard DC blocking caps in the feedback...
Garden variety electrolytics. I just needed to replace a bad
cap in the feedback network of a non-paralled high freq. driver
(also 3886).
I'm going to use servos because the circuit actually auto nulls any DC in the output, so as to eliminate (or almost eliminate) any DC regardless of temperature.
Caps in the feedback loop only render the chip offsets to unity gain by blocking DC though the grounding part of the divider.
cap in the feedback network of a non-paralled high freq. driver
(also 3886).
I'm going to use servos because the circuit actually auto nulls any DC in the output, so as to eliminate (or almost eliminate) any DC regardless of temperature.
Caps in the feedback loop only render the chip offsets to unity gain by blocking DC though the grounding part of the divider.
Oh yea, Like Tiltedhalo mentioned just a few posts back:
"The problem with non servo'ed designs are stability no matter how close your parts are matched minute differences will become apparent as differences in current sharing resulting in waste even if as little as 5W or as great as 50w, and sonic pit falls.
By the way Roland used LM3886's in a inverting mode to set idle DC offset."
Wasted power and extraneous heat are the big issues. Running cooler means more available power before the thermal protection engages. Oscillation (instability) is the worst threat, followed by stray offsets.
"The problem with non servo'ed designs are stability no matter how close your parts are matched minute differences will become apparent as differences in current sharing resulting in waste even if as little as 5W or as great as 50w, and sonic pit falls.
By the way Roland used LM3886's in a inverting mode to set idle DC offset."
Wasted power and extraneous heat are the big issues. Running cooler means more available power before the thermal protection engages. Oscillation (instability) is the worst threat, followed by stray offsets.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.