Further thoughts on this, it seems possible to investigate how well a match is over the spectrum, conducting matches would probably be very labor consuming.
Soon, I don't know where you are going and what will come out of trying to make this more complicated.
I told you in post 251 that I was working with DC and you read my post262 where I confirmed DC.
I have also told you that I measured the bridge output on the PCB, not on some test apparatus.
I told you in post 251 that I was working with DC and you read my post262 where I confirmed DC.
I have also told you that I measured the bridge output on the PCB, not on some test apparatus.
AndrewT,
I am curios about your concern about the "additional impedances of parasitics" of board traces. If we just talk DC, these would not show up in bridge measurements.
What is your opinion about the feedback to the LM318 breaking the bridge balance? I think it does at the DC level. But I certainly would like to understand more at the AC level and the "additional impedances of parasitics".
I am curios about your concern about the "additional impedances of parasitics" of board traces. If we just talk DC, these would not show up in bridge measurements.
What is your opinion about the feedback to the LM318 breaking the bridge balance? I think it does at the DC level. But I certainly would like to understand more at the AC level and the "additional impedances of parasitics".
Hi,
See below sim for the effect of a detuned Howland in the MyRef. Detuning is by setting one resistor to -2%, 0% and +2% of its nominal value, giving the pump different output impedances, including negative.
Input voltage and hence output voltage is a sine as it must be with any voltage amplifier, but to demonstrate the effect I choose a non-correlated square load current for an equivalent of about 10Ohms load.
The output voltage of the first opamp is shown, it is the control voltage for the Howland. As expected, this voltage tracks the current but on top of it a partial "copy" of output voltage can be found. Also we find that even exact resistors (and flawless single pole opamps) doesn't make the voltage part fully disappear as the Howland isn't perfect even with exact resistors, for many of the reasons explained in the paper I had linked to earlier.
Note that with different load currents the control voltage will keep tracking the current whereas the voltage part will stay constant, detemined only by the output impedance of the pump. In the shown setup the point when both are equal in the order of magnitude is reached when the load impedance would be around 70 Ohms which also is the level of output impedance of the Howland in the detuned cases (-70 Ohms and +70 Ohms, dependent on which direction the resistor is wrong).
This means that at points of very high speaker impedance a key point of the MyRef -- making the control and error voltages a function of current rather than voltage -- is missed by a significant margin when the resistors are not matched to at least 1%. Still the degradation from the ideal is not too bad, and resistor matching beyond 0.1% or fine-tuning the howland for parasitics to be a truly ideal pump doesn't have any merits.
Like Andrew I would encourage (true blind) listening tests to confirm any audible differences from a given (known) amount of detuning the Howland down to about (+/-) 100 Ohms output impedance but personally I would not expect any to show up, this will probably be too small a change.
See below sim for the effect of a detuned Howland in the MyRef. Detuning is by setting one resistor to -2%, 0% and +2% of its nominal value, giving the pump different output impedances, including negative.
Input voltage and hence output voltage is a sine as it must be with any voltage amplifier, but to demonstrate the effect I choose a non-correlated square load current for an equivalent of about 10Ohms load.
The output voltage of the first opamp is shown, it is the control voltage for the Howland. As expected, this voltage tracks the current but on top of it a partial "copy" of output voltage can be found. Also we find that even exact resistors (and flawless single pole opamps) doesn't make the voltage part fully disappear as the Howland isn't perfect even with exact resistors, for many of the reasons explained in the paper I had linked to earlier.
Note that with different load currents the control voltage will keep tracking the current whereas the voltage part will stay constant, detemined only by the output impedance of the pump. In the shown setup the point when both are equal in the order of magnitude is reached when the load impedance would be around 70 Ohms which also is the level of output impedance of the Howland in the detuned cases (-70 Ohms and +70 Ohms, dependent on which direction the resistor is wrong).
This means that at points of very high speaker impedance a key point of the MyRef -- making the control and error voltages a function of current rather than voltage -- is missed by a significant margin when the resistors are not matched to at least 1%. Still the degradation from the ideal is not too bad, and resistor matching beyond 0.1% or fine-tuning the howland for parasitics to be a truly ideal pump doesn't have any merits.
Like Andrew I would encourage (true blind) listening tests to confirm any audible differences from a given (known) amount of detuning the Howland down to about (+/-) 100 Ohms output impedance but personally I would not expect any to show up, this will probably be too small a change.
Attachments
Last edited:
Hey Dario, Where are the latest top choices for the FE - the Mouser BOM or a shared Google Docs folder (that I can't find)? I'm about to put in the PRPs in R104/204 and I have some 68Rs for C101/201, but the Caddocks will probably win out. Four Elna Cerafine caps (100uf/35V) are on hand for C102/202.
Also, you mentioned several more items will be surface mount on the final PCB. Will it be possible and/or worth the effort to adapt those changes to the beta boards?
BTW- I found some excellent new test music that contains about everything one might want, hi to low - soft to "rattle the rafters"
Amazon.com: Copland: Symphony No. 3 - Symphony for Organ & Orchestra: Aaron Copland, Leonard Bernstein, New York Philharmonic, E. Power Biggs: Music
Also, you mentioned several more items will be surface mount on the final PCB. Will it be possible and/or worth the effort to adapt those changes to the beta boards?
BTW- I found some excellent new test music that contains about everything one might want, hi to low - soft to "rattle the rafters"
Amazon.com: Copland: Symphony No. 3 - Symphony for Organ & Orchestra: Aaron Copland, Leonard Bernstein, New York Philharmonic, E. Power Biggs: Music
Thanks - I'll try to wrap my mind around this, haven't fully digested it. The one thing I'd like to simulate is the output Z with very slightly mismatched (say 0.1%) resistors in the pump network. The closed-form expression for Zo was in your earlier reference (from Apex), but here we don't actually know the exact open-loop gain of the chipamp, just an approximate gain number from the datasheet for the LM3886.
I'll plug in the LM1875 model for the chipamp and play around with it later.
Hi Bob,
https://docs.google.com/spreadsheet/ccc?key=0AuqKRBuaKtzYdHhjME11dnRUVkNjTndaSDdPOWJjTEE#gid=0
You should have the link to the complete folder in one of earliest emails to beta testsers
PRPs, while fine, are a bit too harsh in that positions I largely prefer KOA PR1 and best Caddocks MP915.
The four Cerafines have a very nice timbre and dynamic but they introduce, I discovered later, a tiny bit of HF harshness.
Now I prefer FGs there.
Probably not.
The through hole conterparts, though, are way more costly than the best SMDs I've selected.
Thanks D, I have that link but thought there might be a separate FE folder somewhere - not so. I'll compare what I have and order the new items for review.
Have you had the opportunity to listen to the True Copper caps for C13. From what I can see they are ~$30 US and would be the most expensive part of the amp.
I'll have to sharpen my component extraction skills for the parts transfer when the new board is available.
Have you had the opportunity to listen to the True Copper caps for C13. From what I can see they are ~$30 US and would be the most expensive part of the amp.
I'll have to sharpen my component extraction skills for the parts transfer when the new board is available.
I may be jumping the gun, no actual burn-in yet, but the Elna Cerafines in C102/202 sound great. My first impression is like moving 8 to 10 rows closer to the stage - resulting in the perception of the band being wider and the instruments more easily defined. All these upgrades are producing both improved punch where it belongs and overall smoothness without sounding forced or fake. Musical indeed!
I'm starting to feel like someone being led around by the Pied Piper, but Dario is truly making some valid and productive recommendations. I will try the other Caddocks and the FGs next after these have marinated.
I'm starting to feel like someone being led around by the Pied Piper, but Dario is truly making some valid and productive recommendations. I will try the other Caddocks and the FGs next after these have marinated.
Not yet, in this moment Intertechnik has production problems and most values are not available...
I'll have to sharpen my component extraction skills for the parts transfer when the new board is available.
Apart the obvious desoldering station, I usually put more solder on the joint and while heating it I gently pull the component but you can do that only with parts not too sensitive to heat.
Thanks.
If you want to speed up burn-in I suggest you to buy the Isotek Full system Enhancer.
One or two plays of the second track and quite all components are full burned-in.
Just have to mention - I'm hearing lots of wind instrument players taking breaths.... SWEET!!
Did you installed them with sockets? If so try to pull them out...(the amp must be powered off, obviously
)Hey Dario,
I bit the bullet and bought two True Copper 1.2uF caps. No non-stock warnings yet.
Have you tried the low temp Chip-Quick solder.
Not sure about the Enhancer CD. Sounds like snake oil to me.
Seems like a few plays of The Rite of Spring and a cover of a couple "Weather Report" disks would do the same thing.
I ran out of sockets but they are in the Mouser order placed today.
I bit the bullet and bought two True Copper 1.2uF caps. No non-stock warnings yet.
Have you tried the low temp Chip-Quick solder.
Not sure about the Enhancer CD. Sounds like snake oil to me.
Seems like a few plays of The Rite of Spring and a cover of a couple "Weather Report" disks would do the same thing.I ran out of sockets but they are in the Mouser order placed today.
I know... a little birdie said to me...
I did the same this morning...
No but I've tried the desoldering alloy, incredibly good.
You could use it to desolder also th parts.
Not sure about the Enhancer CD. Sounds like snake oil to me.
I was skeptical too... I've read about it on the italian forum VideoHiFI
One forumer I trust recommended it and I choose to try.
It works and it's effective.
Fine
Amazon.com: XLO/Reference Test & Burn In: Xlo Reference Recordings, Various Artists: MusicHey Dario,
I bit the bullet and bought two True Copper 1.2uF caps. No non-stock warnings yet.
Have you tried the low temp Chip-Quick solder.
Not sure about the Enhancer CD. Sounds like snake oil to me.
I ran out of sockets but they are in the Mouser order placed today.
Suburra told me it works too but when he tried the Isotek he had more immediate and directly audible results.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.