let's tweak this amp!!!!!

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Small changes may help this circuit to work even better

Questo circuito e molto buonno fratello.

A good idea to increase the power capacity of that BD639..... install something bigger there...and that can have better gain...better than 180 of DC gain.

I think the good transistor for this position may be 2SC2238 or 2SC2328...have to check those ones data sheets, to discover wich one is for audio driver applications.

Alike the Lamborghini Murcielago, this amplifier has a very nice performance.

Increasing the voltage and re-adjusting bias in the provided trimpot, will produce a very good increase in the power....but you will need one more output pair....more than 200 watts RMS over 4 ohms and undistorted power will not be difficult to reach.

Good idea is to remove all that output protection...aaaaagh!...it sounds very bad! and killed dinamics to protect output devices....grumpf!..not a good thing...no one wants a Ferrari with a limiting speed device!

Those enormous capacitors used in the output...from Colector to Base.... hummm...they avoid oscilations...but avoid also good sound....go reducing them and monitoring R22 with the Finger-O-meter, as this one turning hot will means that hi frequencies, from oscilations, are going to ground passing inside this resistor....searching this point of oscilation you will know the mininum possible capacitance to work there...increase one more step to be safe (normal capacitor values standard...those steps).

For instance, as an example...if you reduced using the capacitor standard steps and the amplifier started to oscilate using 100 picofarads..... this may represent that will not oscilate using 120 picofarads...one step more...use 150 picofarads.

Those Darlingtons use to produce some harshing sound...as they start to oscilate when driving high audio frequencies...... this is the reason people install big capacitors over them....this will avoid harshing when driving those High level and High frequency sounds.....but will produce a permanent distortion, even when not driving those high frequencies...so...it is some medicine that will kill the one is sick..... overdosis.

Good luck.

arrivederci

Carlo
 
thanks everybody for the suggestions!!!
Carlo,
can i use 2sc2240 or bd139 in place of bc639?
and the input pair? the bc556 can work with rails of +/-60v?
i would use this amp as subwoofer amp, then suond quality is not my primary target, i would only obtain more power with double output pair and higher supply rails...
and... can you explain how to paralleling output darlinghton? ex: i need to doubling also power resistors(1ohm)and collector-base capacitors?
Grazie fratello!!!
 
Yes, i think you can use the selection type C of the BC556 transistor.

Selezione C capiche leggionari?

Use the selection type C, understood Roman Soldier....but it will work near the upper limit...a little bit dangerous...maybe you can find something more adequated...having more voltage capabilities..to be more safe to you...2N5551 and 2N5401 are units hardly used and can fit your needs too.

The output darlington may have a small resistor in series with the base....both 4 bases will need those 10 ohms resistors.... all emitters must go to througth emitter resistors.... each one twice the schematic value or something near that.

The capacitance from colector to base must be reduced, and it is more safe to use one to each transistor.

As you will use to bass...to subwoofer, the annoying treble will not bother you.

Yes, the BD139 C (atention to selection C) will fit your needs as voltage amplification...it is very good in that position.

Good luck,

Ciao

Carlo dei Danesi
 

Attachments

  • bc556c.jpg
    bc556c.jpg
    68.2 KB · Views: 1,000
Re: Yes, i think you can use the selection type C of the BC556 transistor.

destroyer X said:

Yes, the BD139 C (atention to selection C) will fit your needs as voltage amplification...it is very good in that position.

It's a bad choice, as the amp will get +/60v, it has to withstand 120v, way too much for BD139, it will blow... (Max rating of 80v)
I guess some nice Hitachi or Sanyo video amplifier will be good here.
Or simply a MJE340, often used here.

Mike
 
Yes!...Michael is rigth...bad choice..... as maximum theoric swing will go to 120

Peak to peak....in the practical work...real work it will not go...but will blow because this one is for 80 volts swing.

Yeah...it is better to try another one.

Those higher voltage transistors are found internally to PC supplies...they will fit your needs without problem.

regards,

Carlos
 
ok..ok...lets see if i understand...
-for the bc639 i will use a mje340
-for the input pair i will use 2n5401, but can i use 2sc2240(i just have 2 or 3 of them)?
-for paralleling the bdx66/67 they need to have theirs emiter resistors(1ohm 9w, for each), a base-collector capacitor each, and all the bases must have a 10ohm resistor in series.

please explain me pratically and simply...
thanks for the great help...
 
You can't use 2sc2240, that's npn but you need pnp, that would be 2sa970. The bc556 has max rating of 65v, you might keep it but is dangerous close. If you don't have 2sa970 use the 2n5401 it has rating > 140v.
You might increase r9 to ~8k to avoid overheating of D1.
I think you should add a resistor (~500ohms) between R1 & C1, otherwise C1 can't work correct.
The other changes are correct. You will need a small heatsink for the mje340, it has to dissipate ~1w, too much for a naked to126.
What are you going to use as output devices ? The bdx67b is too weak... (max rating of 100v)

Mike
 
Yes, that`s it....you can use your 2SC2240 as it can work under 120 volts

And your amplifier will not swing 120 volts peak to peak.

Maybe 100 or 110...but never the maximum as there are losts from colector to emitter and also reduction in voltage using zeners in the input.

Yes.... you will like this amplifier to bass....really, it is only perfect for basses.

C2 can be increased...if too big will produce a nice effect...R20 can be reduced to 2,2 ohms to load this amplifier into high frequencies...as you will not use those frequencies it will be loaded only if some oscilation starts naturally or your source provides you some sample frequency not filtered.

Remove R14, R16, T5 and D2....also do the same with the negative rail protective units...they are a hell bad to the sound.

You already have 270 ohms in series with the output base....replace them for 2 units of 270 ohms of even lower value..you can reduce to 100 ohms without problems...those units will feed the darlington base.

regards,

Carlos
 
There are a number of things that should be done to this amp before even building it as it is.

1) R5, R9 and D1
R9 and D1 make up a stabilised voltage source, which in itself is a good idea, since it limits the maximum C-E voltage of the input stage transistors to the zener voltage of D1 plus 40V, i.e. 49 V.
Unfortunately, there are two problems. First, since the input stage is a long tailed pair, R5 is supposed to emulate a current source - but in order to get a tail current of 2.5mA as used in the design, R5 has to be a rather low 3.3k - so the long tailed pair does not really have a very long tail!
The solution is to increase the zener voltage of D1, and, if needed, use input stage transistors with a higher maximum Vce. 2N5401 is often used for that purpose, but they are not ideal - you want something with high gain and high Ft, which the 2N5401 only partially satisfies, but transistors with those characteristics usually have a max Vce of 60V or less. The 2N5401 usually comes up due to it's high max. Vce. One solution is to look at some japanese parts, from the 2SAxxx family. Another is to use something like the MPS8599.
With a MPS8599, D1 could be a 27V zener, R5 can then be increased to 10k, and R9 decreased to 1.8k. Using a higher voltage zener has the additional advantage of making the zener less noisy. Adding a cap of about 100u across the zener will kill the remaining noise.
Other mods could be done to the input stage, most notably, constant gm degeneration, but this would work best with a current source in the tail instead of the simple resistorm because the resistor value would again need to be relatively low.

2) R10, R11, C5
The bootstrapping arrangement is fine, but there is really no reason to use different values for R10 and R11. Normally, equal value provides good compromise between the effective impedance of the bootstrap 'current source' and best usage of the available capacitance of C5. You want 1/(2 x PI x (R10 || R11) x C5) to be well below the bottom of the audio band, usually well below 1Hz!
One way is to increase C5, the other to balance R10, R11. With the same C5, you could try 1.8k for R11, 2.2k for R10. Also, use at least 64V rating for C5 - best ESR performance from standard electrolytics at a given capacitance usually happens between the 50V and 100V ratings. You may also safely increase the value.
Finally, put a reverse polarised diode across the series combination of R10, R11, i.e. from C of T3 to + rail. This is not a major issue but it will prevent T7 from going into deep saturation on a positive clip, which may produce 'sticking' or 'overhang' during positive clip recovery.

3) R7, R8, P1, T3
This sort of arrangement for biac surrent adjust is positively dangerous. If the wiper of P1 ever goes open, bias current becomes 'infinite'. This may happen simply while P1 is being adjusted! Link together the wiper of P1 and the junction of P1 and R7. Make R8 = 330ohms. Also, put a good foil cap (0.1 - 1u) across C-E of T3. T3 should, of course, be on the heatsink between the output darlingtons.

4) R14-19, T5, T6, D2, D3
The output protection is unlikely to protect the output as it is set up now. It limits the current to 16A, which is only the maximum current for the transistor, which can be used only with very small voltages across it. The 'protection' makes no provision for any kind of SOA tracking, so it is likely the transistors will fail for low and reactive impedances. BDX66/67 are not completely useless as far as SOA is concerned, but with only one pair and potentially well over 40V across the transistors at peak load current, they are almost guaranteed to fail for any kind of difficult load and short circuit, with the protection doing nothing. This kind of protection does have the advantage that it has no fold-back effect which makes protections generally sound horrible when activated, but as it is set up, you could just as well remove it completely.
This brings us to R18 and R19 which are unusually high - for this sort of application, one would expect something between 0.1 and 0.47 ohms. 1 ohm does increase the probability of the output stage surviving, but IMHO not nearly enough, and it introduces some power loss. It is also possible it makes the bias current overcompensate for temperature rise. It is sure that 50mA is not the optimal bias for this amp with Re's of 1 ohm, as mentioned in literature by D. Self, but I won't go into this here.
I would definitely rework the protection, which would include lowering the emitter resistors, moving the diodes elsewhere, in order to at least make a single-slope SOA tracking arrangement.

5) T7, T8
BDX66/67 were quite good darlingtons in their day, as far as darlington packs can be good. Usually they have base leak resistors which make it very difficult to properly bias them.
The problem with using these transistors now, is that they are only made by 2 manufacturers now - one is semelab, which supports things like army and navy, hence these would be very expensive, the other is one no-one ever heared of - which makes it likely these will be the ones you get. Unfortunately, this kind of replacement made somewhere in the far east tends to have substandard specs, which aggravates problem (5) above.
I would look for replacements, and even use a discrete darlington connected stage instead. If you still want integrated darlington packs, my favorite would be the 2SB1570/2SD2401 pair. Although on paper they have a 12A current rating, this is misleading because the SOA performance is WAY ahead of the BDX66/67, so is the Ft, power dissipation, and Vce max. They also don't have the base bleeed resistors and present the exception to the rule of problematic darlington packs. They also don't have a 'catch' diode (antiparallel with C-E), so you need to add it onthe outside, a 1N5407 will do nicely, although a faster diode is even better. Incidentally, catch diodes are MANDATORY if any kind of inductive laod is used on amps with overcurrent protection. These usually have no problem with oscillation given base resistors, which are already included in the schematic, and may even alow you to reduce or remove C7 and C8, which will lower HF distortion.

6) T4
Keep in mind that the BC639 is essentially a BD139 in a smaller and less cool-able package!!! Performance will be equal but the amp will be more robust, since you can put a heatsink on a BD139.
The BC639 will dissipate about 0.4W even at rest in this design, which will make it very hot. A BC139 is recomended even with no other mods, with a heatsink!
The BC639/BD139 is however practically at it's maximum Vce rating here, I would recomend a japanese replacement, from the 2SC line, preferably in a TO220 case. I am not aware of any kind of EU or US replacement that would fit as nicely. BD139 is almost ideal but for the Vce max limitation.
With a heatsink it may be prudent to increase somewhat the stage current, by reducing R10 and R11 to 1.8k or even 1.5k each. With 2x1.8k T4 will dissipate 0.41W at rest, and with 2x 1.5k T4 will dissipate about 0.5W so a relatively small heatsink will be OK.
Keep in mind that increasing the current increases stage gain, as well as slew rate, and in this case, also the output DC offset. The distortion will go down somewhat, but IMHO the changes should not be enough to require a change in the frequency compensation caps.

7) There should be a resistor inserted between R1 and C1 in the input line! I am guessing about 2k or so would be used here, but it depends on what you will drive the amp with.
 
Flavio88 said:
ok..ok...lets see if i understand...
-for the bc639 i will use a mje340
-for the input pair i will use 2n5401, but can i use 2sc2240(i just have 2 or 3 of them)?
-for paralleling the bdx66/67 they need to have theirs emiter resistors(1ohm 9w, for each), a base-collector capacitor each, and all the bases must have a 10ohm resistor in series.

please explain me pratically and simply...
thanks for the great help...

Paralleling output transistors also requires a corresponding current increase in the VAS stage. This requires either an increase in input stage tail current or a different Q4, and may or may not have repercussions to stability.
BTW the amp loses about 12W in the two emitter resistors at full power, 8 ohms, twice that at 4 ohms, which makes a signifficant loss, and also is beyond the resistor rating for sustained full power at 4 ohms.
As I pointed out in the previous post, there are a few issues with this amp that need to be addressed, before trying to make a 'turbo' version.

One more thing: C4. DON'T use a 4V rating :) a 63 or 100V 1000uF cap is not that big any more, and since the audio signal current goes through it, you really want a good performance.
An even better option would be a bipolar cap, but that would be very large. Another alternative is a low ESR cap as used for core voltage supplies on some PC motherboards, this will have a 6.3, 10 or 16V rating, but for that sort of cap it is OK. Use a higher rating if you can find one. There should be plenty of old ones where the caps are still OK :). In order not to destroy the cap in the event the output blows, you may want to put a protection network considting of two antiparallel diodes like the 1N4148 in parallel to the cap, but this is not critical (the voltage across the cap will be very low in normal operation) as long as you remember that if the output stage ever blows, you MUST replace this cap. This is often missed when repairs are made!
 
MikeB said:
Hi Carlos, in input diffamp the voltage is not doubled, only one half supply plus the input swing. This full swing thing is only valid for vas and output.
Mike

Mike, this is of course true for normal operation. Experience shows me, however, that one should look at possible failure modes before considering what rating to use. In particular, I have seen MANY amps where the NFB DC blocking cap and the input pair were damaged when the output devices fail, in the case of this amp, it would happen if the top device failed short, which is not an unlikely scenario. The simplest way to prevent it is to protect the DC blocking cap so the NFB divider always limits the maximum voltage that can appear at the LTP bases.
In this design, the LTP transistors should ideally sustain one supply rail plus zener voltage of D1 as Vce. The BC556 are already very close, the power supply could only be increased by about 10V, assuming all other potential problems are addressed as well.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.