Hi Zenmod,
but: f5 for tweeters NEED low DC offset...
I'm still thinking about using 2 F3's, firing with a balanced pre, and employ a servo to keep both outputs on the same dc level... ;-)
Dirk
but: f5 for tweeters NEED low DC offset...
I'm still thinking about using 2 F3's, firing with a balanced pre, and employ a servo to keep both outputs on the same dc level... ;-)
Dirk
we discuss that previously , I think ...... 😉
DC coupling of tweets isn't good , at least in my book - from two reasons which you already know - you need servo and you need anticlick ;
anticlick with speaker protection is must , at least for anyone having expensive/rare speakers
blablah ..... anyway - you already know that ....
I'm just puzzled about question - is series cap to tweet less evil than servo ?
be nice , and report your experience , when you try it 😉
Hello, on compensation of opamps ("control theory") there should be a lot of material on the internet. sorry I dont have a tutorial at hand, but try Analog, Embedded Processing, Semiconductor Company, Texas Instruments, they have a great section on application notes, including a rather large text (600+ pages) on opamps, and I believe they talk about compensation as well.
And yes, simple circuits that work well have beauty in them, I think; I admit I also do have the NTCs and all the rest, but I was interested in doing it for the fun of it and so I did....
;-)
Please keep in mind that the "normal" DC servo connection for "audio amplifiers" is back to the inverting node of the front end's differential pair. There is no differential pair in the F5.
_-_-bear
_-_-bear
two solutions and probably many others.
1.) take the non-inverting DC servo output to the source of the input FETs, just like NP does with the 50r feedback resistors.
2.) use an inverting DC servo and take the output to the input FET gate.
Returning a DC voltage to the input JFet gate would not be a good idea since the input is not AC coupled.
Andrew, did you say if you have built an F5 yet? 😀
_-_-
Andrew, did you say if you have built an F5 yet? 😀
_-_-
F5 with servos (schematic)
Sorry for the delay, as announced last week here is the schematic for the F5 with servos.
Couple comments:
>> Two LM2904s are needed, since their max VCC is 26V, and I had to get the lower close to the source potential of the n/channel power MOS. It is important to use an opamp that will allow for the inputs to be able to go down to its negative rail (this opamp has PNP inputs so no problem).
>> In this schematic, the output overcurrent protection is still in there, but the NTC is not (no longer needed....)
>> The lower circuit uses a TL431 to create a temp-compensated reference voltage, against which the voltage on the source resistor of the n-channel power MOS is compared and regulated accordingly - no real magic here. The upper circuit compares the output voltage to ground, and regulates the p-channel power MOS accordingly - no real magic here either.
>> R41 and R42 can be replaced with a trimpot to adjust the bias current.
>> Note C96 - this is to "disable" Q11 working as a grounded-base amplifier (the original circuit is pretty immune to (50Hz) noise on the rails, the better the gain between the two halves is matched the lower the noise. this circuit injects a lot of noise without C96 - with a regulated PSU you probably won't need it)
>> The caps on the rails are just for local bypassing - tried the amplifier with a switchmode lab supply (and long cables) and it oscillated nicely at 300kHz..... the caps made that disappear immediately . I put +/-20V but it doesnt really matter.
hope you like it!
still need some time to build a couple channels with the original circuit and compare the sound.....
Sorry for the delay, as announced last week here is the schematic for the F5 with servos.
Couple comments:
>> Two LM2904s are needed, since their max VCC is 26V, and I had to get the lower close to the source potential of the n/channel power MOS. It is important to use an opamp that will allow for the inputs to be able to go down to its negative rail (this opamp has PNP inputs so no problem).
>> In this schematic, the output overcurrent protection is still in there, but the NTC is not (no longer needed....)
>> The lower circuit uses a TL431 to create a temp-compensated reference voltage, against which the voltage on the source resistor of the n-channel power MOS is compared and regulated accordingly - no real magic here. The upper circuit compares the output voltage to ground, and regulates the p-channel power MOS accordingly - no real magic here either.
>> R41 and R42 can be replaced with a trimpot to adjust the bias current.
>> Note C96 - this is to "disable" Q11 working as a grounded-base amplifier (the original circuit is pretty immune to (50Hz) noise on the rails, the better the gain between the two halves is matched the lower the noise. this circuit injects a lot of noise without C96 - with a regulated PSU you probably won't need it)
>> The caps on the rails are just for local bypassing - tried the amplifier with a switchmode lab supply (and long cables) and it oscillated nicely at 300kHz..... the caps made that disappear immediately . I put +/-20V but it doesnt really matter.
hope you like it!
still need some time to build a couple channels with the original circuit and compare the sound.....
servo circuit
Once again -- very impressive Hessner.
Where did you mount the components -- on a seperate board?
Steve
Once again -- very impressive Hessner.
Where did you mount the components -- on a seperate board?
Steve
Q: U1A and U2A are unused?
_-_-
Btw, just thinking aloud, it would seem to be sufficient to keep one side manually adjusted to bias up and then servo adjusting the other to make 0vDC offset?
_-_-
Btw, just thinking aloud, it would seem to be sufficient to keep one side manually adjusted to bias up and then servo adjusting the other to make 0vDC offset?
No one said it did.
It obviates the need to readjust the bias manually for zero DC offset - valuable in some situations for some people.
It remains to be seen what is reported about its effect on sound if any.
_-_-bear
It obviates the need to readjust the bias manually for zero DC offset - valuable in some situations for some people.
It remains to be seen what is reported about its effect on sound if any.
_-_-bear
OK, sounds like a lot of added complication to an F5 with no real benefit for DIY , a real plus if you are manufacturing ...
regards,
regards,
Last edited:
I like the servo solution. But how does it behave on "Switch on"? Does it need an output relais???
Dirk
Dirk
Hi Patrick,
yes! Package arrived at my parents, but unfortunately I had no time to visit them...
Everything looks fine (as reported by my father...) ;-)
Thanks. Will come back to you as soon as I took a look!!!
Dirk
yes! Package arrived at my parents, but unfortunately I had no time to visit them...
Everything looks fine (as reported by my father...) ;-)
Thanks. Will come back to you as soon as I took a look!!!
Dirk
Hi all, thanks for your comments.
Regarding the sound - built a stereo amp and it did sound very nice to me (I posted my impressions earlier in this thread). Will try and build two "original" amps and report on the differences ( if any....). But since the impact of those bipolars are really small, and the short circuit protection effectively represents a collector capacitor as well (I removed those), the difference should be minimal.
Regarding construction - my PCB has all of this stuff on one PCB, I will post pictures probably tomorrow.
Regarding differences - I did not plan on building more than what I would like to have for private use, so volume manufacturing is no concern for me.... but since MOSFETs can age, and the threshold voltage can drift, I did have a little concern with output offset increasing over time, and bias current changing over time, which will not happen here. The bias current is constant, cold or warm....
Regarding the sound - built a stereo amp and it did sound very nice to me (I posted my impressions earlier in this thread). Will try and build two "original" amps and report on the differences ( if any....). But since the impact of those bipolars are really small, and the short circuit protection effectively represents a collector capacitor as well (I removed those), the difference should be minimal.
Regarding construction - my PCB has all of this stuff on one PCB, I will post pictures probably tomorrow.
Regarding differences - I did not plan on building more than what I would like to have for private use, so volume manufacturing is no concern for me.... but since MOSFETs can age, and the threshold voltage can drift, I did have a little concern with output offset increasing over time, and bias current changing over time, which will not happen here. The bias current is constant, cold or warm....
sorry, one more: at switch-on, the servos will settle with their time constant so theoretically in a few seconds. I found them to settle in less than 500ms in all cases. My amplifiers do have an output relay, with turn-on delay and additional protection against DC at the output (if you have a relay, thats easy to add)
Would I be save without the output relais? What is the maximum peak voltage on turn on?
(Of course, it depends on how big the influence of the servo circuit may be...)
Do you plan a pcb for sale / to share? Output relais plus delay circuit on board???
Regs, Dirk
(Of course, it depends on how big the influence of the servo circuit may be...)
Do you plan a pcb for sale / to share? Output relais plus delay circuit on board???
Regs, Dirk
I can't say for sure about hesener's implementation, but people have reported some sonic differences when the protection circuit was present vs. not.
I do not like output relays at all. Ymmv.
My approach would be to set up the bias components so that the amp was biased and stable without the servo operating, and then permit the servo to adjust any offsets that subsequently occur.
In the F5 I have experience with the turn on offset was minimal. No thump.
Wonder what others have found?
_-_-bear
I do not like output relays at all. Ymmv.
My approach would be to set up the bias components so that the amp was biased and stable without the servo operating, and then permit the servo to adjust any offsets that subsequently occur.
In the F5 I have experience with the turn on offset was minimal. No thump.
Wonder what others have found?
_-_-bear