AMB Laboratories DIY Audio

[Thago]

Would it be possible to use expensive resistors (ultimately something like Vishay 'naked' Z-foil) for the Rs resistors (these are in line with the signal), and cheaper ones for the Rp-resistors (these provide a path towards ground)? Or would the mix with lower quality resistors inhibit the sound improvement that is desired by using the more expensive Rs-resistors?

[linux-works]

if you build such a thing, I'd be happy to do some tests on it for you.

but honestly, I think you'd be throwing money away.

all the resistors are in the signal path. parallel is, too.

I have not done any tests with higher quality than vishay/dale.

its not clear what any benefit would be, other than bragging rights and keeping your seller in a bigger car and home

but if you do build with such parts, I wouldn't mind doing a test to SEE if there is any mesured benefit.

my gut feeling is: its not worth overspending on. put better money toward good buffering and a clean psu for the buffer.

[Thago]

Hi

Thank you for replying! It would be very nice if you could perform an objective test. Unfortunately, I don't plan to visit the States the next months, but you are welcome in Europe (Belgium)
I agree that all resistors are in the signal path, including the parallel ones. But the way I understand the schedule it is the following: when the relay bridges Rp, than Rp is outside the signal path; when the relay connects Rp, part of the signal is diverted to ground, and will not appear at the output. If this is correct, then the quality of the Rp resistors will not influence the signal quality, is that right? Probably the tolerance of Rp would matter, and needs to be in the same order as Rs in order to keep signal coherent between L and R, and between the different steps of the attenuator. But is should be possible to find 0.% Rp's for much less money than Z-foils.

Something I am wondering about in the resistor value calculator, is the value of Zload I need to use. I plan to combine the delta-1 (+ delta-2) in front of a buffer output (Pass B1 or DCB1), to create a passive pre. Zatt would be 25kohm, but what is Zload then in this situation? Also 25kohm?

http://www.amb.org/audio/delta1/delta1_ ... e2_big.png

[amb]

Something I am wondering about in the resistor value calculator, is the value of Zload I need to use. I plan to combine the delta-1 (+ delta-2) in front of a buffer output (Pass B1 or DCB1), to create a passive pre. Zatt would be 25kohm, but what is Zload then in this situation? Also 25kohm?

In the original First Watt B1 schematic, R100, R101, R200 and R201 are 1M ohms. If you retain these resistors then that would be the input Z of the buffer. Use that for your Zload for calculating Rt (i.e., the default value if you're using Zatt=25K ohms and 8 relays with 0.5dB steps).

If you do not populate those resistors then you can treat Zload as "very high" and just use Rt = Zatt.

[linux-works]

Hi

Thank you for replying! It would be very nice if you could perform an objective test. Unfortunately, I don't plan to visit the States the next months, but you are welcome in Europe (Belgium)

you could mail the completed boards and I would test them similarly to how I did boinger's boards. I have an automated data collector that can step thru every single db value and measure the real db attenuated. I later plan to do freq response checks (adding another inner loop). this might be the most telling of the tests.

so, if you are game to spend the money on high end resistors, I'll offer to run some number gathering on them.

my guess, though, is that it won't justify its cost in terms of numbers. but I really dont' know as I've not built with expensive parts yet. the 30 cents per vishay is about as rich as I have ever used, on this project

I agree that all resistors are in the signal path, including the parallel ones. But the way I understand the schedule it is the following: when the relay bridges Rp, than Rp is outside the signal path; when the relay connects Rp, part of the signal is diverted to ground, and will not appear at the output. If this is correct, then the quality of the Rp resistors will not influence the signal quality, is that right?

when a relay is in a position to short out the series R and 'let go' of the parallel R, neither of those are 'in the circuit'

BUT, in the other case where the relay is not like the above, then both are in the circuit. how could one be and not the other?

if you really want to get high end tweaky (lol) then you can skip all the 'bad steps' that have the most amount of stages in play. each relay is a bit and that is a stage. the stage is either fully bypassed and out of the equation or its in-play. numbers like 7 that have 4+2+1 (the bottom 3 stages all in non-bypass) are 'more evil' - if you want to see it like that - since they take more stages to do their work. the saintly number 8 takes only 1 stage to be in play and all the rest are at rest you start to get more and more evil as you get to 15, then you are saintly again at 16.

and you SEE this if you run passive. with no buffer, the more evil you get, the more noise you really do hear.

and so, you should only allow yourself to step to 0, 2, 4, 8, 16, 32, 64 db and that's it.

(LOL!)

but if you have a buffer directly (very very directly, measured in millimeters) after then the evilness is nullified to a great extent this is why I insist on running with an active buffer after my d1 attens. only exception is when the d1 is co-located inside an amp and that amp is the buffer.

I guess what I'm saying is that the R's on the d1 board are not going to change your sound, at least I don't think they will. but spending your time on getting the best electrical (logical, short wiring) connection between the d1 and its buffer; and also the psu for the buffer, THAT will be your payback for money spent.

[linux-works]

if you really want to know which resistor type (brand, model) works best, you'll have to try the experiment yourself.

that's the only way to know, anyway. the only way to know for sure.

some may be more capacitive and some may be more inductive. or exhibit freq response changes differently when in series or parallel.

there's also the effect of the pc board and the relays. that also adds stuff to the network.

above 1 mhz, I bet all this will matter (lol). at audio, I seriously doubt it. but if you want to know, you'll have to setup tests, get good test gear and do some before/after sweeps. THEN you can know that your expensive resistors are worth it.

I would NOT - simply NOT just throw money at this and assume you have gotton better sound. more expensive and exclusive parts don't, by themselves, mean a thing if you don't place them in context. I'd avoid cheap 10% carbon resistors (lol) but I don't think there's honestly a need to spend more than the dale/vishay style parts. they are well toleranced, very quiet and well respected parts. they don't cost all that much and they are available (usually) in all the values you need for the d1.

[Thago]

Thank you very much for the comprehensive argumentation, I think I begin to understand what you mean. The price tag would indeed be very substantial, for something that probably makes no sense at all, or the extra value is at least questionable. I will use very good standard parts. With the money I save, I can build a very good power supply for the buffer, and even then there may be something left over for some other project too. Thanks for the advice!

[Thago]

In the original First Watt B1 schematic, R100, R101, R200 and R201 are 1M ohms. If you retain these resistors then that would be the input Z of the buffer. Use that for your Zload for calculating Rt (i.e., the default value if you're using Zatt=25K ohms and 8 relays with 0.5dB steps).

If you do not populate those resistors then you can treat Zload as "very high" and just use Rt = Zatt.

Thank you for clarifying this!