PeeCeeBee

VSSA T03

It seems to me that all TH boards in this thread could probably work without gate resisitors. Only Borys's pcb with two pairs of HexFets must have gate resisitors. Gate resistors are necessary with multiple Fet output pairs. If you remember the generic Hitachi schematic that I sent to you (with one pair of output transistors), there was no gate resistors. All pcbs featured in this thread are compact, with short pcb tracks and good layout with topographic placement of components that go from input to output. Therefore there should be no problems in using one output pair without gate resistors. When you repair the boards used for "antenna experiment" you can replace gate resistors with jumpers and see if bandwidth is more extended compared with resistor version and if stability is exemplary.

Hi Ivanlukic
I have made ​​these small changes and there are reasons that have been made them by LC.
Really not need gate resistors Only 100pf colector to ground .Comp cap 5,6pf but when pull out 100pf and no comp cap i have oscilation with 1K and 47 Ω but with 2,2k and 100Ω no osc at all .All these with bd 139 and 140 .
I made some changes to the board Jason and working very stable
on 47 Ω 120mv to other 121mv
on 470 Ω 870 mv to to other 875 mv
1,6V total on gate without gate resistor
only 100 pf col to ground 10pf comp- CCS 2N5462 - 1k and 47Ω- offset 1-3mv positive.
I hope to not be a problem by Μr Shaan and Jason
 

Attachments

  • 2.GIF
    2.GIF
    54.7 KB · Views: 625
Last edited:
VSSA T03

Hi Ivanlukic
I have made ​​these small changes and there are reasons that have been made them by LC.
Really not need gate resistors Only 100pf colector to ground .Comp cap 5,6pf but when pull out 100pf and no comp cap i have oscilation with 1K and 47 Ω but with 2,2k and 100Ω no osc at all .All these with bd 139 and 140 .
I made some changes to the board Jason and working very stable
on 47 Ω 120mv to other 121mv
on 470 Ω 870 mv to to other 875 mv
1,6V total on gate without gate resistor
only 100 pf col to ground 10pf comp- CCS 2N5462 - 1k and 47Ω- offset 1-3mv positive.
I hope to not be a problem by Μr Shaan and Jason

Small change on CCS
 

Attachments

  • 4-1.GIF
    4-1.GIF
    54.9 KB · Views: 554
Last edited:
Small change on CCS
Oh No...
No CCS please :D
Shaan will not happy with it

read this:
Hi fellow diy-ers. :wave:

Important:- Visitors/posters are welcome to modify any of the layouts according to their need and/or add totally new designs. However, if you want to use CCS or Multilayer Layout or Parallel capacitors in the feedback network or Resistive bias spreader then please post your VSSA experiment and results in the original VSSA thread. Link >Here<. This thread's topic is single layer printed circuit board designs based on the VSSA schematic without the above mentioned accessories, shown in one of the pictures attached below. Thanks for your co-operation.

:)

the VSSA thread is moved, not in solid state anymore :eek:
 
Hi!

Why do you need input cap? In the original VSSA schematic there isn't any.

Tyimo
Reason is the same as any amplifier circuit. You need it mainly to block any DC offset in the input signal. DC at the input would appear at the output amplified by the gain of the amplifier. In most cases this is a bad, for the speakers and for other reasons.

Some amp/preamp combinations work direct coupled, but usually that assumes good control of the grounds and good offset calibration in the entire signal chain.
 
It seems to me that all TH boards in this thread could probably work without gate resisitors.
Agreed, but I can't think of a way to prove that it will work with any component combinations we have been discussing here. For example, I really look forward to JKuetemann's finding on those driver transistors, and building a set of boards with those...

... All pcbs featured in this thread are compact, with short pcb tracks and good layout with topographic placement of components that go from input to output...
There would actually be a significant benefit to removing those resistors completely on my board. There is a tradeoff that had to be made: shortest track v. single signal layer v. best component placement... something to be discussed if other people are still interested in another rev of my board.

... When you repair the boards used for "antenna experiment" you can replace gate resistors with jumpers and see if bandwidth is more extended compared with resistor version and if stability is exemplary.
I am waiting for parts from Tayda in Thailand (I automatically just clicked on the cheapest shipping option, now I wish I had the parts in hand... :rolleyes:)

Full circle - as you know, I don't mind experimenting... it is sort-of part of my hobby, if I burn up a board or two, so be it, I have spares to listen to. Even though this may be a stable circuit with any reasonable care, we should also say that it does not take too much to make it oscillate, and the result could be some damaged components.
 
Hi Ivanlukic
I have made ​​these small changes and there are reasons that have been made them by LC.
Really not need gate resistors Only 100pf colector to ground .Comp cap 5,6pf but when pull out 100pf and no comp cap i have oscilation with 1K and 47 Ω but with 2,2k and 100Ω no osc at all .All these with bd 139 and 140 .
I made some changes to the board Jason and working very stable
on 47 Ω 120mv to other 121mv
on 470 Ω 870 mv to to other 875 mv
1,6V total on gate without gate resistor
only 100 pf col to ground 10pf comp- CCS 2N5462 - 1k and 47Ω- offset 1-3mv positive.
I hope to not be a problem by Μr Shaan and Jason

nikosokey,

This is very useful information concerning TH version of VSSA.
 
Really not need gate resistors Only 100pf colector to ground .Comp cap 5,6pf but when pull out 100pf and no comp cap i have oscilation with 1K and 47 Ω but with 2,2k and 100Ω no osc at all .All these with bd 139 and 140.
Very nice!, and thanks for posting your results.

However, I am not sure I agree with the conclusion. My logic is this: If using 1K in the feedback can cause oscillation with fairly slow drivers (BD139/140), and if everything else was the same, then we should be a bit careful when using other, faster drivers.

If eliminating the gate resistor means adding a 100pf cap, I am not sure what is gained. If you want to leave out the compensation caps completely, and at the same time have to add 100pF elsewhere, and at the cost of a more complicated layout, is there really an advantage?

On my board, a compensation cap on the driver transistors has almost zero impact on the layout. Adding a 100 pF close to the collector means bringing a new ground track into the middle of the board, using either a jumper, or a very long track. Clearly this makes the layout more complex, even if the gate resistor is eliminated. It also makes it more difficult to allow for a larger footprint for VAS heatsinks, which may be necessary for anyone who wants to raise the current maximum supply voltage.

....

I would like to have a board that follows Shaan's basic idea to have a VSSA that anyone can build with parts available almost anywhere, etch by himself, or use a professionally made one like I do. I think this is what makes Shaan's approach worthwhile.

If I wanted to make a VSSA just for myself, I would use CRD's, two-sided layout, supply split at the VAS stage and regulated, Exicon or Alfets, etc. I still think about a design for a double-die Exicon... :cool:. But, how many people would want to build one???
 
There would actually be a significant benefit to removing those resistors completely on my board. There is a tradeoff that had to be made: shortest track v. single signal layer v. best component placement... something to be discussed if other people are still interested in another rev of my board.

Even though this may be a stable circuit with any reasonable care, we should also say that it does not take too much to make it oscillate, and the result could be some damaged components.

Gate resistors should be used by diy-ers that do not like to experiment too much, that want their pcb to work immediately without any problems, from the first power-up. This is simple preventive measure.
 
Reason is the same as any amplifier circuit. You need it mainly to block any DC offset in the input signal. DC at the input would appear at the output amplified by the gain of the amplifier. In most cases this is a bad, for the speakers and for other reasons.
Ofcourse! But with DC free output ( capacitor coupled DAC or preamp) it needn't.

Greets.
Tyimo
 
Ofcourse! But with DC free output ( capacitor coupled DAC or preamp) it needn't.

Greets.
Tyimo
My guess is, LC left it out to save space. The through-hole boards are bigger to start with, so leaving it in is the more correct approach, because putting a jumper across the leads (if you don't want to use it), is easier than an off board cap with flying leads.

edit:

See this post in the VSSA thread about what can happen even with "DC free direct coupled signals".

http://www.diyaudio.com/forums/vend...lateral-mosfet-amplifier-167.html#post3539760

Don't get me wrong, I am absolutely in favor of modifying just about anything... amplifiers, cars, coffee pots... but I have also learned rule #1 of basic modding... don't do this with your DD (daily driver)!:D
 
Last edited:
My logic is this: If using 1K in the feedback can cause oscillation with fairly slow drivers (BD139/140), and if everything else was the same, then we should be a bit careful when using other, faster drivers.

If I understand nikosokey finds, with BD pair, all we need for stability is 2K2 feedback resisitor and 100R gain setting resistor.

In that case we do not need neither of the three:

1. gate resistors
2. VAS compensation cap
3. colector to ground cap

which in fact means unlimited choice of VAS transistors from ubiquitous BD pair to something better.

That is consistent with some of Shaan's comments on stability of the circuit and with theory of current feedback amps. With CFB amps there is no constant gain/bandwidth product. Bandwidth is not dependent on ratio of feedback and gain setting resisitors but on value of feedback resistor. This is very straightforward way of limiting bandwidth and influencing stability. Why should we be limited to 1K/47R values? Lower values improve noise performance of the ciruit but are not beneficial for stability. I've seen even lower values in circuit similar to VSSA (820R/16R) but also with some additional compensations.
 
My guess is, LC left it out to save space. The through-hole boards are bigger to start with, so leaving it in is the more correct approach, because putting a jumper across the leads (if you don't want to use it), is easier than an off board cap with flying leads.
O.K. I see.

Is there any rule of thumb for PSU sizing for Class AB amp?

Greets:
Tyimo
 
O.K. I see.

Is there any rule of thumb for PSU sizing for Class AB amp?

Greets:
Tyimo
Not really. There are various considerations. The type of load, 8 ohm or 4 ohm, peak power v. continuous power, and the amount of acceptable noise and ripple.

I originally sized my power supply for low-to-moderate power into 8 ohms.

250 VA transformer 25V AC, rms, dual secondary
Schottky diode rectifier w. minimal heatsink
3x 4700uF per rail, in CLCRC passive filter configuration
+/-35V DC nominal, 37V max rail voltage
3A fast blow glass fuses in power supply rails

Since then I decided to use a power supply with an active filter, capacitance multiplier type, from a circuit posted by MrEvil here in the power supply section. I did that because I gradually realized that this amplifier needs much cleaner power supply rails for best performance than I first thought, especially without CCS.

I posted some addition info about that here, but not tested yet:

Linear Capacitance Multiplier Supply