Coils update: Coilcraft
initial testing of the Coilcraft {they have 5-6 18mH models} I ordered the 56 Ohm-rated DCR model, they are rated at 56DCR .5mHz SRF no shield.
I sampled say 10 of the 100x I received today, typical readings are steady at around 47 Ohms R give or take {probably the same number of windings?}
and most of the samples ran around 18.7mH but there ARE some hints there could be 2-3 grades for matching purposes, so FAR.. They could be unwound under a few pair of reading glasses worn at same time. DCR would fall maybe almost -2 Ohms? 45? The tolerance and calibration of the LCR meter of course is also in question. I have 90 more coils to measure and another measuring device coming to re_check all my values. Plus I did not take any notes so I have 100x coils to grade. Lower than published DCR is probably good.
As old and somewhat experienced as I am I cannot visualize what 18.7 vs 18mH means to this filter {how much slack is there by design} -- like thinking of Morse code as dots and dashes {that doesn't work} or thinking of what 18mH would do to a 8-ohm speaker crossover network {that obviously does not work}. I can unwind and scrape and flux and solder, or reject the coils but probably they are within specs, so I am inclined to not be a pain to anyone.
The Coilcraft look well-built for what they are. In Qty 100 they ran, even with shipping, well under a dollar, maybe 70-80 cents US Ea.. I'd like to try them, wish I could this eve.
initial testing of the Coilcraft {they have 5-6 18mH models} I ordered the 56 Ohm-rated DCR model, they are rated at 56DCR .5mHz SRF no shield.
I sampled say 10 of the 100x I received today, typical readings are steady at around 47 Ohms R give or take {probably the same number of windings?}
and most of the samples ran around 18.7mH but there ARE some hints there could be 2-3 grades for matching purposes, so FAR.. They could be unwound under a few pair of reading glasses worn at same time. DCR would fall maybe almost -2 Ohms? 45? The tolerance and calibration of the LCR meter of course is also in question. I have 90 more coils to measure and another measuring device coming to re_check all my values. Plus I did not take any notes so I have 100x coils to grade. Lower than published DCR is probably good.
As old and somewhat experienced as I am I cannot visualize what 18.7 vs 18mH means to this filter {how much slack is there by design} -- like thinking of Morse code as dots and dashes {that doesn't work} or thinking of what 18mH would do to a 8-ohm speaker crossover network {that obviously does not work}. I can unwind and scrape and flux and solder, or reject the coils but probably they are within specs, so I am inclined to not be a pain to anyone.
The Coilcraft look well-built for what they are. In Qty 100 they ran, even with shipping, well under a dollar, maybe 70-80 cents US Ea.. I'd like to try them, wish I could this eve.
I just simmed the passive input section of JG's filter in LTSpice with 15mH, 18mH and 22mH as the inductor values. Its showing a minor (0.1dB) lift at 20kHz which gets slightly bigger as the inductor value goes up - this is caused by the inductor ringing. Nothing much to worry about. The design could be made less sensitive to inductor value by including a damping resistor across the coil. From simulation I reckon 47k is about right.
<edit> I just noticed curryman's circuit is a little different from the JG original, no series 40R so my result doesn't apply to that circuit, only JG's.
<edit> I just noticed curryman's circuit is a little different from the JG original, no series 40R so my result doesn't apply to that circuit, only JG's.
Last edited:
The Fastron DS shows a max of 17R for the DC resistance. When you ask for the sensitivity wrt the DC resistance, do you mean you want to add more series resistance with a resistor or you'd like to check with different DCR inductors? If the inductor's DCR changes then its SRF will also change, this latter effect will dominate.
I'm curious about one aspect of your schematic (post #528) Daniel - you have two 4n7s in parallel directly on the output of the DAC. What's the purpose of them without any series resistance from the DAC side? Seems to me they have the potential of making the output opamps unstable when there's no series resistance. JG puts a 240R series resistor in the schematic for his filter.
I'm curious about one aspect of your schematic (post #528) Daniel - you have two 4n7s in parallel directly on the output of the DAC. What's the purpose of them without any series resistance from the DAC side? Seems to me they have the potential of making the output opamps unstable when there's no series resistance. JG puts a 240R series resistor in the schematic for his filter.
Oh that's very confusing for an idiot like me that those components are explicit in the schematic 
Simulating with a pure resistor for output impedance isn't a very good idea because in practice, for an opamp the Zout looks inductive, due to reduction in feedback as the frequency climbs. I'm getting the feeling that I may have opened up a can of worms here...
Simulating with a pure resistor for output impedance isn't a very good idea because in practice, for an opamp the Zout looks inductive, due to reduction in feedback as the frequency climbs. I'm getting the feeling that I may have opened up a can of worms here...
Last edited:
The two parallel C in my schematic are optional/alternative (different packages). Originally I used a SMD PPS cap (e.g. Pana EHCU) but there have been several requests for WIMA FKP foil caps (through hole) so finaly both can be used (alternatively)
We really should not discuss the output impedance of opamps here. The Filter has proven to work very well and everyone is happy
On the other hand: at what frequency the inductive part becomes relevant? And what would be the effect on the filter response? mhh... I'll play with the sim a bit this evening
We really should not discuss the output impedance of opamps here. The Filter has proven to work very well and everyone is happy
On the other hand: at what frequency the inductive part becomes relevant? And what would be the effect on the filter response? mhh... I'll play with the sim a bit this evening
Yes I've heard very positive feedback from a listener to the filter, so yes its definitely working.
To model an opamp output there's the open loop output impedance (which might indeed be around 200R or so for a CMOS opamp) at the unity gain frequency (probably 10MHz or so) and then this goes down with decreasing frequency so at 10kHz it might be 0.2ohm. The overall effect I'll try with an inductor for Zout in parallel with the 240R and also taking out the 40R now I'm slowly getting up to speed
<edit> The new inductor turns out around 3uH with my estimates and the plot is here - problematic is the ringing above 1MHz.
To model an opamp output there's the open loop output impedance (which might indeed be around 200R or so for a CMOS opamp) at the unity gain frequency (probably 10MHz or so) and then this goes down with decreasing frequency so at 10kHz it might be 0.2ohm. The overall effect I'll try with an inductor for Zout in parallel with the 240R and also taking out the 40R now I'm slowly getting up to speed
<edit> The new inductor turns out around 3uH with my estimates and the plot is here - problematic is the ringing above 1MHz.
Attachments
Last edited:
OK then assuming the 240R is a real resistance in series with the opamp output here is how it looks (47k inductor damping resistor still in circuit) :
Inductor peaking is about 1.2dB at 70kHz quite benign. Second plot shows how it looks without the 47k - almost 3dB peaking.
Inductor peaking is about 1.2dB at 70kHz quite benign. Second plot shows how it looks without the 47k - almost 3dB peaking.
Attachments
Last edited:
Just for the sake of completeness, here's what happens to the FR when the inductor (with damping) is changed down to 15mH and up to 22mH - zero audible change to the audio band response - well under 0.1dB difference at 20kHz.
<edit> Please note this does not mean you can change the nominal inductor value and get no difference - this is just to show the effects of 20%-ish tolerance on the 18mH value specified.
<edit> Please note this does not mean you can change the nominal inductor value and get no difference - this is just to show the effects of 20%-ish tolerance on the 18mH value specified.
Attachments
Last edited:
Coilcraft DCR 47-17 = 30 40R-30=10R...or not
So if Coilcraft DCR stays at 47, minus the 17 of the 07MFG Fastron, that's 30 Ohms... and the JG schematic calls for 40 just before the 07MFG coil as I recall.. not meaning to put you out, would I just plunk in 10 ohms series before the Coilcraft -- or just ES9023 DAC -> 150R {or 240} 4.7n to ground, 47DCR/18.7MH w/ 47K parallel, 18.##K to ground...etc? I feel like it must be "close".
Someone uttered OP AMP? Once I decided I must try this, I had impure thoughts about tubes, and wondered what of all the tube circuits is like the buffer stage? [I have 6088, 6021, 6n16b-v 6n1p, 2p, 3p, 6p...12Au7, 6sn7, and the cute Mickey Mouse-like single triode: 7193...] . I was about ready to make the Broskie Unbalancer with 6n16b-v x 4 when I discovered I like the ES9023 which is already single-ended, defeating the purpose of the Unbalancer, though one differential cap-free input could simply be grounded. I have the PCB but that seems complex for the task {and has excess gain to dump later} Pondering this I think it would be a circuit that would NOT have a series CAP at it's input? And SIMPLE, maybe 12-24-30V plate. {no more complex than PSU for the FETs} No flames please, I just like the glow of a valve. I'll try something off_line "in my spare time".
So if Coilcraft DCR stays at 47, minus the 17 of the 07MFG Fastron, that's 30 Ohms... and the JG schematic calls for 40 just before the 07MFG coil as I recall.. not meaning to put you out, would I just plunk in 10 ohms series before the Coilcraft -- or just ES9023 DAC -> 150R {or 240} 4.7n to ground, 47DCR/18.7MH w/ 47K parallel, 18.##K to ground...etc? I feel like it must be "close".
Someone uttered OP AMP? Once I decided I must try this, I had impure thoughts about tubes, and wondered what of all the tube circuits is like the buffer stage? [I have 6088, 6021, 6n16b-v 6n1p, 2p, 3p, 6p...12Au7, 6sn7, and the cute Mickey Mouse-like single triode: 7193...] . I was about ready to make the Broskie Unbalancer with 6n16b-v x 4 when I discovered I like the ES9023 which is already single-ended, defeating the purpose of the Unbalancer, though one differential cap-free input could simply be grounded. I have the PCB but that seems complex for the task {and has excess gain to dump later} Pondering this I think it would be a circuit that would NOT have a series CAP at it's input? And SIMPLE, maybe 12-24-30V plate. {no more complex than PSU for the FETs} No flames please, I just like the glow of a valve. I'll try something off_line "in my spare time".
Seems you are a bit confused
As a result of the simulations: you don't have to worry at all about the coils DCR and you don't need to add a series resistor there. The 40 Ohm in the original schematic just represents the (estimated) DCR of an arbitrary coil for the initial development of the circuit. However as abraxalito pointed out the influence of a different DCR is almost neglegible.
The same for the 240 Ohm series resitor in the original schematic: it just represents the output impedance of the ES9023. You don't have to add an additional resistor here
And finally the filter does not react too much on a varying inductance. E.g. between 15 and 22mH the influence is quite small. Matching between stereo channels still seems to make sense.
Best regards, Daniel
Confused no more. +47K then PAR
So ES9023 -> 4.7nF to GND to 47K in Parallel with the {matched} coil, 18.9K to GND...470R 150P to GND. Z-in of actual buffer is probably hiding in the above thread. Impure thoughts <deleted>.
Now, a slight reason for my confusion is ODAC uses some R there after the DAC I think 220.
Thanks for setting me straight!!
{The Other Pat}
Seems you are a bit confused
As a result of the simulations: you don't have to worry at all about the coils DCR
-snip-
And finally the filter does not react too much on a varying inductance. E.g. between 15 and 22mH the influence is quite small. Matching between stereo channels still seems to make sense.
Best regards, Daniel
So ES9023 -> 4.7nF to GND to 47K in Parallel with the {matched} coil, 18.9K to GND...470R 150P to GND. Z-in of actual buffer is probably hiding in the above thread. Impure thoughts <deleted>.
Now, a slight reason for my confusion is ODAC uses some R there after the DAC I think 220.
Thanks for setting me straight!!
{The Other Pat}
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.