Guys, I'd like to know, if you prefer se or bal using amps like the LPUHP or the LME-amp. What are the advantages and disadvantages in signal-quality? I also ask myself if a bal source Haß a better signal-quality, when it's used bal or if it doesn't matter, when inv is grounded. Did anybody measured around my issue?
The background to my question:
I use a UCX as source for my LPUHP's, which are se-connected. Because of external impacts I might need to make them balanced and would like to know, what happens inside the circuit and how the result changes. Why I connected them se is because I wanted to avoid parts, which the signal has to pass, so two LME49990 in this case.
Best regards,
Stammheim
The background to my question:
I use a UCX as source for my LPUHP's, which are se-connected. Because of external impacts I might need to make them balanced and would like to know, what happens inside the circuit and how the result changes. Why I connected them se is because I wanted to avoid parts, which the signal has to pass, so two LME49990 in this case.
Best regards,
Stammheim
Last edited:
Amplifier, etc...
"Finally, which amplifier are you using this with? Noise at the output is very odd, and indicates something might be adrift with the amp itself. Is it prone to oscillating? I've tried the NTD1 with dozens of amplifiers at this point, and have never heard anything at all at the output. I'm also curious why it would have struggled with 1V of common mode DC voltage at the input. An amplifier with a proper differential input stage should be fine with a few volts of common mode DC. It's diff mode DC that causes problems, and that should have only been a few mV worst case."
The amplifier in question is a DIY NC-400 stereo amp. To be sure, the level of noise is very low, inaudible at the listening position, and one might not notice it with an amplifier which itself has some noise. Within about a foot of the tweeter on a quiet evening one can hear the course noise floor I have. I have experimented with different analog filter cut offs on various Twisted Pear I/V with the Buffalo DAC, and have found this same noise at the amplifier outputs anytime one relaxes the filter at all. As an example, the original analog filter on the IVY-III results in no noise at the amp output, but if one relaxes the caps from 15 nF to 8.2 nF, one can start to hear the course noise... and the stock 2.2 nF caps on the Legato 3 filter circuit definitely let this noise appear.
BTW, I allowed for about 100 hours with the polymer caps and still had about a volt of DC one each output, still too much for me to happy with.
The 3 µF MusiCaps are sounding quite good, certainly no discernible bass loss (into my amp with 100K input impedance). A little more burn in time is necessary for ultimate evaluation, but so far it is sounding good.
t is easy enough for me to experiment a bit with adding an analog filter, so I shall probably try it soon...
"Finally, which amplifier are you using this with? Noise at the output is very odd, and indicates something might be adrift with the amp itself. Is it prone to oscillating? I've tried the NTD1 with dozens of amplifiers at this point, and have never heard anything at all at the output. I'm also curious why it would have struggled with 1V of common mode DC voltage at the input. An amplifier with a proper differential input stage should be fine with a few volts of common mode DC. It's diff mode DC that causes problems, and that should have only been a few mV worst case."
The amplifier in question is a DIY NC-400 stereo amp. To be sure, the level of noise is very low, inaudible at the listening position, and one might not notice it with an amplifier which itself has some noise. Within about a foot of the tweeter on a quiet evening one can hear the course noise floor I have. I have experimented with different analog filter cut offs on various Twisted Pear I/V with the Buffalo DAC, and have found this same noise at the amplifier outputs anytime one relaxes the filter at all. As an example, the original analog filter on the IVY-III results in no noise at the amp output, but if one relaxes the caps from 15 nF to 8.2 nF, one can start to hear the course noise... and the stock 2.2 nF caps on the Legato 3 filter circuit definitely let this noise appear.
BTW, I allowed for about 100 hours with the polymer caps and still had about a volt of DC one each output, still too much for me to happy with.
The 3 µF MusiCaps are sounding quite good, certainly no discernible bass loss (into my amp with 100K input impedance). A little more burn in time is necessary for ultimate evaluation, but so far it is sounding good.
t is easy enough for me to experiment a bit with adding an analog filter, so I shall probably try it soon...
Hi Sean,
Please see below two PDF documents with the top layers printed to scale.
I've verified these here, but you will need to make sure you print them to scale, so check them against the boards to make sure they match.
All holes are intended to be 4-40 machine screws, but you can use whatever fits.
The NTD1 itself has 8 critical holes (the 8 around the output stage) and 2 optional holes (the holes in the upper left and right corners).
The PSU has 9 critical holes by the regulators and 3 optional holes by the AC inputs.
It's important to snug these screws up, but not to over-tighten them. it's fine to see a slight dimple in the PCB around the screw, but it should barely be visible. If you really crank down on these, you risk deforming the PCB and possibly stressing adjacent components, so proceed lightly.
The heat sinking mostly works through the large surface area, so a lot of pressure is not necessary. A thin gap-pad is ideal here , anywhere from 0.3-0.7mm thickness is ideal.
Best of luck, and let me know how it turns out!
Regards,
Owen
Please see below two PDF documents with the top layers printed to scale.
I've verified these here, but you will need to make sure you print them to scale, so check them against the boards to make sure they match.
All holes are intended to be 4-40 machine screws, but you can use whatever fits.
The NTD1 itself has 8 critical holes (the 8 around the output stage) and 2 optional holes (the holes in the upper left and right corners).
The PSU has 9 critical holes by the regulators and 3 optional holes by the AC inputs.
It's important to snug these screws up, but not to over-tighten them. it's fine to see a slight dimple in the PCB around the screw, but it should barely be visible. If you really crank down on these, you risk deforming the PCB and possibly stressing adjacent components, so proceed lightly.
The heat sinking mostly works through the large surface area, so a lot of pressure is not necessary. A thin gap-pad is ideal here , anywhere from 0.3-0.7mm thickness is ideal.
Best of luck, and let me know how it turns out!
Regards,
Owen
Hi Owen,
Thank you for all of your devotion to DIYer.
I almost finished my prototype of ES9018 with NTD1 V3 IV stage.
The sound is very(x3) nice...
I like this NTD1 rather than my first IV stage.
My former IV stage was single op-amp (AD797) circuit from sabre reference board.
I don't know how can I express it's sound because my poor english.
Regards,
Sean
Thank you for all of your devotion to DIYer.
I almost finished my prototype of ES9018 with NTD1 V3 IV stage.
The sound is very(x3) nice...
I like this NTD1 rather than my first IV stage.
My former IV stage was single op-amp (AD797) circuit from sabre reference board.
I don't know how can I express it's sound because my poor english.
Regards,
Sean
Attachments
This is great to know, thanks Owen! 100R and 50R are standard values in the Riedon PFS35 series, so it seems to make the most sense to simply go with those at $3.76 each.
I'm not sure I'd trust myself to take a Dremel to a bunch of $30 resistors anyway.
When you get the chance, my only other question was whether you have a recommended reduced value for R5, R6, R15, and R16 in this case. My guess was 9 or 10k, assuming the pot stays at 5k.
DCX2496 uppgrade.
Soldered it up, but lacked 4 resistors
R13 R15 R19 R21 R25 R27 R31 R33 R37 R39 R41 R43 R47 R49 R53 R55 Susumu RES 1.0K OHM 1/8W .1% 0805 SMD RG2012P-102-B-T5 754-RG2012P-102-B-T5 12
Should be 16.
Found some 2k that i can use. A little more noise. Less gain and larger load.
Haven't tried it yet.
The groundplane eats heat by the way. A little difficult to solder the 1uFs. But I managed.
I am looking forward to try it out. Really nice board!
Soldered it up, but lacked 4 resistors
R13 R15 R19 R21 R25 R27 R31 R33 R37 R39 R41 R43 R47 R49 R53 R55 Susumu RES 1.0K OHM 1/8W .1% 0805 SMD RG2012P-102-B-T5 754-RG2012P-102-B-T5 12
Should be 16.
Found some 2k that i can use. A little more noise. Less gain and larger load.
Haven't tried it yet.
The groundplane eats heat by the way. A little difficult to solder the 1uFs. But I managed.
I am looking forward to try it out. Really nice board!
Hey Owen,
A thought bubble ... have you ever been tempted to build up 4 of the LT3042 pcbs and hook each of the boards up in series to NTD1? Costly and more bulky than the NTD1 PSU though.
Actually looking at the datasheet now, is the LT3042 really able to be set for 20V output as mentioned in the wiki page write up? Looks like it has 15V max output.
Chris
A thought bubble ... have you ever been tempted to build up 4 of the LT3042 pcbs and hook each of the boards up in series to NTD1? Costly and more bulky than the NTD1 PSU though.
Actually looking at the datasheet now, is the LT3042 really able to be set for 20V output as mentioned in the wiki page write up? Looks like it has 15V max output.
Chris
Last edited:
Owen,
I see a number (282) in the paid column but the two PC boards have not been shipped yet. Did you you receive the Paypal payment?
Please let me know if i need to do anything on my side.
In the next few Weeks I will be moving to my winter location. My current location is
Matthew Levin
3243 88th St. Apt B305
East Elmhurst, NY 11369
Thanks,
Mattdude0 282
I see a number (282) in the paid column but the two PC boards have not been shipped yet. Did you you receive the Paypal payment?
Please let me know if i need to do anything on my side.
In the next few Weeks I will be moving to my winter location. My current location is
Matthew Levin
3243 88th St. Apt B305
East Elmhurst, NY 11369
Thanks,
Mattdude0 282
Owen,
Quick question... I just assembled one of your LT3042 parallel regulator boards. While in the datasheet, I noticed that it is specified for a lower limit output voltage of 0 volts and I didn't see anything in the datasheet that indicated that the parallel application limited how low one could go in voltage out.
Is the 1.8v you specified incorrect or am I missing something here?
TIA!
Greg in Mississippi
P.S. BTW, if it DOES go down to 0v, I have a great 1.2v application for these!
Quick question... I just assembled one of your LT3042 parallel regulator boards. While in the datasheet, I noticed that it is specified for a lower limit output voltage of 0 volts and I didn't see anything in the datasheet that indicated that the parallel application limited how low one could go in voltage out.
Is the 1.8v you specified incorrect or am I missing something here?
TIA!
Greg in Mississippi
P.S. BTW, if it DOES go down to 0v, I have a great 1.2v application for these!
Hi Greg,
I think I mixed up the output and input voltages when I was writing the specs for the LT3042 regulator boards. I will fix this in the wiki.
You can indeed set the output right down to 0V, so no problem if you need 1.2V.
I also wrongly indicated up to 20V output when in reality output voltage is limited to 15V.
Regards,
Owen
I think I mixed up the output and input voltages when I was writing the specs for the LT3042 regulator boards. I will fix this in the wiki.
You can indeed set the output right down to 0V, so no problem if you need 1.2V.
I also wrongly indicated up to 20V output when in reality output voltage is limited to 15V.
Regards,
Owen
Hey Owen,
Does this necessitate a change in the plan of running the low-gain NTD1 and BAL-BAL at +/-18V from two LT3042 boards, or is it still in the realm of possibility with adequate heat-sinking?
Assuming this'll still work, I had a couple other questions come up while planning out the build:
Do you see any issue with stacking two 90k ohm resistors in parallel on the boards to get 45k resistance for 18V output? Looks to be impossible to find 45k 0805 resistors in stock.
- Are two 50VA 18VAC dual secondary transformers sufficient for this set up?
How would you recommend running the wires between the LT3042s, NTD1, and BAL-BAL? Separate wires from the LT3042s to each, or daisy-chained between the NTD1 and BAL-BAL?
Do you have a recommended input impedance for the BAL-BAL, running the 190R output impedance of the NTD1 into it?
The boards are awesome, by the way! I'll add some pictures to the thread soon.
Assuming the following current draws per power supply rail, I ran through a few power calculations based on voltages available in the Signal Transformer LP series.
NTD1 (45R/98R): 200mA (per leg)
BAL-BAL: 530mA (max, per rail)
I: 0.2+0.53 = 0.73A
C: 5600*2 = 0.0112F
f: 60Hz
1/2 ripple amplitude: 0.5*0.73/(2*60*0.0112) = 0.272v
10V
Vac: 10*0.9 = 9.0
Vpeak: 9*1.414 = 12.73
Vrect: 12.73-(2*0.550) = 11.63
Vripple: 11.63-(2*0.272) = 11.09v
15V
Vac: 15*0.9 = 13.5
Vpeak: 13.5*1.414 = 19.09
Vrect: 19.09-(2*0.550) = 17.99
Vripple: 17.99-(2*0.272) = 17.45v
17V
Vac: 17*0.9 = 15.3
Vpeak: 15.3*1.414 = 21.63
Vrect: 21.63-(2*0.550) = 20.53
Vripple: 20.53-(2*0.272) = 19.99v
Based on the LT3042 datasheet and the comments a few posts back, I'm going to guess that there's no way to safely run the output at 18v. 18v regulated DC output would also require the 17V transformer above, which comes in at or above the 20v max input of the 3042.
So, if that's the case, would you recommend adjusting the NTD1 resistors to run the rails at +/-15v and further increasing the gain of the BAL-BAL to compensate, or should I get two additional LT3042 boards (four total), set them 9v, and run each pair in series to get +/-18v?
If I do go the route of wiring them in series, is there any issue with still using two transformers (one per channel), each powering two LT3042 boards?
And to add a quick thought to my question in the post above about voltage set resistors on the LT3042, would you recommend paralleling two resistors to get the exact value required, or using a single resistor that is close, say within 100ohms, but not exact?
NTD1 (45R/98R): 200mA (per leg)
BAL-BAL: 530mA (max, per rail)
I: 0.2+0.53 = 0.73A
C: 5600*2 = 0.0112F
f: 60Hz
1/2 ripple amplitude: 0.5*0.73/(2*60*0.0112) = 0.272v
10V
Vac: 10*0.9 = 9.0
Vpeak: 9*1.414 = 12.73
Vrect: 12.73-(2*0.550) = 11.63
Vripple: 11.63-(2*0.272) = 11.09v
15V
Vac: 15*0.9 = 13.5
Vpeak: 13.5*1.414 = 19.09
Vrect: 19.09-(2*0.550) = 17.99
Vripple: 17.99-(2*0.272) = 17.45v
17V
Vac: 17*0.9 = 15.3
Vpeak: 15.3*1.414 = 21.63
Vrect: 21.63-(2*0.550) = 20.53
Vripple: 20.53-(2*0.272) = 19.99v
Based on the LT3042 datasheet and the comments a few posts back, I'm going to guess that there's no way to safely run the output at 18v. 18v regulated DC output would also require the 17V transformer above, which comes in at or above the 20v max input of the 3042.
So, if that's the case, would you recommend adjusting the NTD1 resistors to run the rails at +/-15v and further increasing the gain of the BAL-BAL to compensate, or should I get two additional LT3042 boards (four total), set them 9v, and run each pair in series to get +/-18v?
If I do go the route of wiring them in series, is there any issue with still using two transformers (one per channel), each powering two LT3042 boards?
And to add a quick thought to my question in the post above about voltage set resistors on the LT3042, would you recommend paralleling two resistors to get the exact value required, or using a single resistor that is close, say within 100ohms, but not exact?
The maximum output voltage of a mains transformer is:
Vout = Vmains/Rated Primary Voltage * Vsecondary * transformer regulation
if you have a 230:15+15Vac 8% regulation transformer and run it on a 237Vac supply,
then the output voltage is:
237/230 * 15 * 1.08 = 16.69Vac
But that is not the maximum voltage.
The maximum is determined by the maximum supply voltage. You need to ask your supplier what their maximum supply voltage is.
In Europe the maximum is 253Vac.
The Maximum voltage from that 15Vac transfomer is :
253/230 * 15 * 1.08 * manufacturing tolerance = 17.82Vac ±tolerance.
Now calculate the maximum smoothing capacitor voltage and you arrive at:
17.82*sqrt(2) - diode bridge drop, when delivering no current ~ 24.2Vdc
way above your 19.99Vdc
Vout = Vmains/Rated Primary Voltage * Vsecondary * transformer regulation
if you have a 230:15+15Vac 8% regulation transformer and run it on a 237Vac supply,
then the output voltage is:
237/230 * 15 * 1.08 = 16.69Vac
But that is not the maximum voltage.
The maximum is determined by the maximum supply voltage. You need to ask your supplier what their maximum supply voltage is.
In Europe the maximum is 253Vac.
The Maximum voltage from that 15Vac transfomer is :
253/230 * 15 * 1.08 * manufacturing tolerance = 17.82Vac ±tolerance.
Now calculate the maximum smoothing capacitor voltage and you arrive at:
17.82*sqrt(2) - diode bridge drop, when delivering no current ~ 24.2Vdc
way above your 19.99Vdc
Thanks Andrew, that's really helpful and brings up an excellent point.
My calculations above were for the minimum voltage that the regulator could potentially see from the transformer. However, when operating the LT3042 close to its maximum output it seems wise to check both the lower and upper bounds of your transformer supply voltage.
My calculations above were for the minimum voltage that the regulator could potentially see from the transformer. However, when operating the LT3042 close to its maximum output it seems wise to check both the lower and upper bounds of your transformer supply voltage.
Yes, I agree.
The lowest output voltage determines whether the regulator enters the "drop out" range. That crucifies regulator performance.
The highest output voltage determines the size of the heatsink and whether the regulator survives.
A Designer must look at both operational modes.
And should work with nominal voltages to see how far from nominal performance specifications the regulator and circuit will deviate when voltages are high, or low.
A Builder copies a competent design.
The lowest output voltage determines whether the regulator enters the "drop out" range. That crucifies regulator performance.
The highest output voltage determines the size of the heatsink and whether the regulator survives.
A Designer must look at both operational modes.
And should work with nominal voltages to see how far from nominal performance specifications the regulator and circuit will deviate when voltages are high, or low.
A Builder copies a competent design.
Last edited: