Andrew,
correct me if I'm wrong but I can't understand how a PCB trace, unless extremely thin, could make a significant difference in the bridge balance.
1cm of 2 Oz 70mil trace has a resistance of 0.00137 Ohms (I used this calculator).
So that same centimeter would represent 0,0000029149% of the 47K resistor...
Solder joints have a resistance around 0,0000000000825 Ohms.
It seems that resistors themselves should be the only important factor, where I'm wrong?
correct me if I'm wrong but I can't understand how a PCB trace, unless extremely thin, could make a significant difference in the bridge balance.
1cm of 2 Oz 70mil trace has a resistance of 0.00137 Ohms (I used this calculator).
So that same centimeter would represent 0,0000029149% of the 47K resistor...
Solder joints have a resistance around 0,0000000000825 Ohms.
It seems that resistors themselves should be the only important factor, where I'm wrong?
The length of the leg of a leaded resistor makes a difference to the balance (output) of a bridge.
I don't know how well your calculations model trace resistances, but the resistance of 1cm of lead does affect the bridge balance.
That is working at DC. If you are working with AC, you then have the additional impedances of parasitics to make the "output" error even worse.
If you go back to post251 para3, you will see I have already confirmed that lead length makes a difference.
I don't know how well your calculations model trace resistances, but the resistance of 1cm of lead does affect the bridge balance.
That is working at DC. If you are working with AC, you then have the additional impedances of parasitics to make the "output" error even worse.
If you go back to post251 para3, you will see I have already confirmed that lead length makes a difference.
Last edited:
So, since a cm of 24awg wire should have a resistance about 0,0008 Ohms this means that a variation of 0.0000017021% Vs the 47K resistors gives a measurable difference?
BTW, while measurable differences can be measured I would hardly trust such differences are audible at all... How much difference you measured with different lenght of wires?
Your objection is founded but after building near 15 monoblocks I hope I would have noted some imbalance between channels.
You reported that one of your modules has negative impendance and the other positive...is it possible that all 15 monoblocks I've built have the same impedance (positive and/or negative)?
Possible, maybe, but not very likely...
Yes please ? The ceramic 22pF, the leg on the R3 side comes out pretty close to the BAV99 at D1
Must I populate C102, C202 ? I don't seem to have bought those. I need to find a fastener for the Aavid To-220 for R3. The slip-on Aavid To-220 sinks are for the LM317TG right ? and the LM337 does not need a sink ?
Must I populate C102, C202 ? I don't seem to have bought those. I need to find a fastener for the Aavid To-220 for R3. The slip-on Aavid To-220 sinks are for the LM317TG right ? and the LM337 does not need a sink ?
Wouldn't the resistor tolerances greatly exceed the trace tolerances? I will read that post again later, but could it be the way the trace run have more effect than the trace length themselves? This would be what I expect if a difference could actually be heard.
Yes, but it's not a problem... they're on the same trace.
Do you have problem soldering those two components?
The initial Mouser code gone out of stock, maybe this is the problem.
I'm not using them, in fact it sounds better without.
Probably in the final version they will be omitted.
I didn't shared with you the BOM?
The slip-on are not needed, I recommended them when I thought LM317 had problems but I was wrong.
You should also have two heatsinks that must be bolted with an M3 screw and bolt.
The hole in MP915 must be enlarged with a screwdriver to hold the M3 screw.
Yes, but it's not a problem... they're on the same trace. My bad, I did not realize that.
Do you have problem soldering those two components? no
The initial Mouser code gone out of stock, maybe this is the problem.
I'm not using them, in fact it sounds better without.
Probably in the final version they will be omitted. Great, I will omit the ones I don't have.
I didn't shared with you the BOM? Yes, I made a mistake
The slip-on are not needed, I recommended them when I thought LM317 had problems but I was wrong.
You should also have two heatsinks that must be bolted with an M3 screw and bolt.
The hole in MP915 must be enlarged with a screwdriver to hold the M3 screw. I had planned to dis-assemble a dead commercial amp for it's heatsinks, I should find some usable hardware too
Do you have problem soldering those two components? no
The initial Mouser code gone out of stock, maybe this is the problem.
I'm not using them, in fact it sounds better without.
Probably in the final version they will be omitted. Great, I will omit the ones I don't have.
I didn't shared with you the BOM? Yes, I made a mistake
The slip-on are not needed, I recommended them when I thought LM317 had problems but I was wrong.
You should also have two heatsinks that must be bolted with an M3 screw and bolt.
The hole in MP915 must be enlarged with a screwdriver to hold the M3 screw. I had planned to dis-assemble a dead commercial amp for it's heatsinks, I should find some usable hardware too
Hi Andrew,
you might check this legendary app note on Howland current pumps : http://apex.cirrus.com/en/pubs/whitePaper/199210-Apex-Versatile_current_source_circuits.pdf
It details on mismatch by finite DC gain of the opamp as well as the frequency dependant mismatch coming from its output impedance characteristics.
However, in the MyRef this is all of little consequence because of the tight feedback wrapped around the pump. Even when the pump's output Z drops to, say, 100Ohms this still results in very little additional output on the 318 on top of what is required by the current transfer function alone because this 100Ohms is effectivly in parallel to the speaker and hardly dominating its impedance.
Note that the biggest mismatch will probably come from the effective output impedance (also vs. frequency) of the 318 which spoils any precision matching of the 22K/47k R's.
Last edited:
When I posted we were part of a group buy using Uriah's hand matched resistors.
The Group Buyers knew they were getting matched resistors. That is because the Howland papers that we had seen up till then all specified accurate matching of the resistors ratios for proper operation of the pump.
I tried to find how resistor tolerance affected those critical ratios. Then asked the question. Does it affect the sound?
I will go and read your new link.
The Group Buyers knew they were getting matched resistors. That is because the Howland papers that we had seen up till then all specified accurate matching of the resistors ratios for proper operation of the pump.
I tried to find how resistor tolerance affected those critical ratios. Then asked the question. Does it affect the sound?
I will go and read your new link.
Have you ever investigated the tolerances around a bridge circuit and how small the errors need to be to yield a good "output" that is believable?
Have you built a bridge?
Have you measured a bridge?
I agree tolerances are important, but it seems that resistor tolerances are greater than trace tolerances provided that you don't use VERY accurate meters to hand pick. I don't think these meters even measure other "additional impedances of parasitics" you mention. Additionally, cables used for measurements have these "additional impedances of parasitics" of their own that effect the measurements too.
To answer your question about bridges, that was done some 30 years or so ago.
But I'd also like your opinion on "Have you ever investigated the tolerances around a bridge circuit and how small the errors need to be beyond which YOU feel that further reduced errors are not audibly of concern?" As you can see this can be subjective.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.