the 200r:15r have been optimised to get best performance from the amp, leave it at those values
I and many other don't agree.
The feedback relies absolutely on the RATIO remaining the same for all signals at any passband frequency and for all levels going through the amp.
The two resistors have a very different dissipation due to the division ratio.
For good RATIO accuracy the temperature variations, including instantaneous peaks, must be kept equal.
This is done by ensuring that the proportion of Pmax for each part of the feedback are kept similar.
It does not matter as much, when the Pinstantaneous/Pmax is <1%
But when Pinst/Pmax approaches 25% many designers warn of the performance implications. Many Designers suggest the maximum dissipation of the feedback resistor should be less than 10% of Pmax.
When playing quiet music/audio and Pinst is very low we have no concerns about the ratio variance due to temperature.
When transients pass, then temperature becomes an issue. That concern is minimised (and measurable) when Pinst/Pmax is kept very low.
5off 1/4W, or 5off 1/2W, resistors with >1W (1/5=200mW) dissipation (not peak) does not meet the temperature variance requirement stated above. The instantaneous peak dissipation for the group of five for a 15.5Vac signal is >2Wpk. Ipk >20mA for each of the five resistors.
Last edited:
I am quite happy with the tradeoffs I have made and the design rules I've adopted Andrew.
Rather than lecture me on my design choices, why don't you post one of your designs up so we can see how you do it.
The sx-Amp more than a match I am sure for any other class A amps in its power range.
Rather than lecture me on my design choices, why don't you post one of your designs up so we can see how you do it.
The sx-Amp more than a match I am sure for any other class A amps in its power range.
There is enough space on this pcb to fit 5 x 1w low temp coefficient metal film resistors as I did and Pinnocchio too...
Fab
I'm sure there would be plenty of space for more Pmax capability.
But it's that the Pinstant/Pmax ratio has not been thought about thoroughly that I was raising and then I get a lame argument about sufficient capability.
It's instantaneous changes in temperature of the resistor element.
The average heat is very low as Bonsai has stated. Thus heat build up inside the case is not a concern.
But it's that the Pinstant/Pmax ratio has not been thought about thoroughly that I was raising and then I get a lame argument about sufficient capability.
It's instantaneous changes in temperature of the resistor element.
The average heat is very low as Bonsai has stated. Thus heat build up inside the case is not a concern.
Mouser has 1K resistors with temperature coefficients as low as 5 ppm/C and 15 Ohm resistors with 15 ppm/C. These are 0.1% resistors and cost around $2 each but they should reduce concerns about power modulating the resistance value. Whether the expense is justified or not is up to the individual builder.
It should be noted that equal temperature changes in resistors do not guarantee equal resistance shifts. The temperature coefficient is specified as being within plus or minus limits about zero, such as +/-5ppm/C. If one resistor is +5ppm and its neighbor -5ppm then raising their temperatures the same number of degrees actually causes maximum resistance divergence. Minimizing the temperature coefficient is the only way to minimize resistance divergence.
It should be noted that equal temperature changes in resistors do not guarantee equal resistance shifts. The temperature coefficient is specified as being within plus or minus limits about zero, such as +/-5ppm/C. If one resistor is +5ppm and its neighbor -5ppm then raising their temperatures the same number of degrees actually causes maximum resistance divergence. Minimizing the temperature coefficient is the only way to minimize resistance divergence.
is one way to minimise resistance variation.
Another way is to minimise the temperature change.
I've been catching up on this thread since I first read it last year. Sounds like fun to build and not overly complicated.
I lucked out and bought 8 each of the ksc/ksa 'E' a few years ago before they were discontinued (curses!!!). Even though I have only 8 of each, I bought extra ksa E grades recently so I can have decent matches with the ksc3503 parts.
Time to put them to very good use! I'll pony up for the ebay boards in a few weeks and start checking my parts inventory against the bom.
I lucked out and bought 8 each of the ksc/ksa 'E' a few years ago before they were discontinued (curses!!!). Even though I have only 8 of each, I bought extra ksa E grades recently so I can have decent matches with the ksc3503 parts.
Time to put them to very good use! I'll pony up for the ebay boards in a few weeks and start checking my parts inventory against the bom.
Sorry, just thinking out loud. We only need bottom foil for etching. However for board house boards, it is very cheap to add the top layer with pads and vias, even jumpers. I don't want to create a fuss. Someone mentioned ordering boards. I thought it was simple to add the top layer. I prefer board house boards but can etch too.
Blessings, Terry
Blessings, Terry
You seem to suggest Bonsai's choice of resistors and their power rating might lead to more THD and/or compression.
Can you give us an idea of the level of THD and/or compression you expect from his evil choices?
Or are there other evils ?
As pointed out, you can buy 5ppm T/C (and lower if you are prepared to pay) resistors from Mouser et al if you are specifically concerned about thermal modulation of the feedback resistor value. I've used Vishay CMF60's in my sx and nx Amplifiers without problems.
Let us know how your build goes - be sure to download the article from my website and put the new updated resistors in.
Your post does not deserve an answer, but I give you one anyway.
I did not say "evil".
I said
Using Bonsai's preferred resistor, the Vishay CMF60, the datasheet indicates that 0.22W causes a temperature rise of 25C in the resistor. If the ambient is an uncomfortable 35C the resistor operates at 60C in this case. This is well within the specifications for the resistor so there is no reliability issue.
CMF60 resistors can be bought with 25ppm temperature coefficient so a 25C rise could produce a resistance shift of 625ppm or 0.0625%. If that is too much shift then a more expensive resistor with better tempco will bring the shift down.
I would rather choose a tightly specified resistor and limit the number of resistors instead of the brute force approach of using a very large number of resistors with the inherent parasitics they would bring.
CMF60 resistors can be bought with 25ppm temperature coefficient so a 25C rise could produce a resistance shift of 625ppm or 0.0625%. If that is too much shift then a more expensive resistor with better tempco will bring the shift down.
I would rather choose a tightly specified resistor and limit the number of resistors instead of the brute force approach of using a very large number of resistors with the inherent parasitics they would bring.
Just identifying the feedback R thermal and voltage sensitivity was a step forward for designers. I think I would go to a tighter TC and sufficient Wattage. THx RNMarsh
Last edited: