Scott,
I haven't followed this thread lately, but the Digikey coils sound even better than the air coils? But the values are not as same as the stock coils though. The values are "wrong" for the stock coils, you think?
Perhaps I ought to try...
Thanks,
Duc
Hi Scott,
I'm modding both the 2x100 and the 4x100 boards. Physical fit aside, would you recommend this inductor for the 4x100 board as well?
Tried adding extra 2 x 2200uF Panasonic FM with the Meanwell 24V, it works. Up it to 15mF of Elna and it croaks. 
Some questions regarding soft-start, ElFishi's schematics in post#377 suggested 47ohms 5W, is this power rating okay?
I'd like to use higher resistance and lower wattage, which increases the charging time, if I use 1kohm then my RC will be ard 15 seconds, I'd prefer to live with that than power surges, and I can use a lower-power resistor for this.
Am I thinking correctly?
Some questions regarding soft-start, ElFishi's schematics in post#377 suggested 47ohms 5W, is this power rating okay?
I'd like to use higher resistance and lower wattage, which increases the charging time, if I use 1kohm then my RC will be ard 15 seconds, I'd prefer to live with that than power surges, and I can use a lower-power resistor for this.
Am I thinking correctly?
Well, got some bad news (for me). Looks like I'm going to have to order another board as I think I fried mine. The soldering iron i was using is way too big for this application and I messed something up. Mine came with a fan on the HS so it never got hot at all, so I decided to remove the fan thinking it'd still be fine. Well it actually did get pretty hot with the fan off (at 32v) so I decided to put it back on. Desoldering the fan leads caused no problem but when I tried to put it back on, it was much harder and I messed something up. May have fried the little chip that the one fan lead goes to... a +5v line I think? anyways the board turned on and the fan spun but now both the power light and the mute (blue) lights were on, and there was no sound. So I went to desolder the fan leads again and the next time i turned it on, still no sound at all and the power light instead of being fairly bright green was a dim yellow. to make things worse after all of this i ended up shorting out 2 of the pins on that chip that the fan lead connects to (+5v with 3 l eads out... i think) and it sparked and then the SMPS started making a weird noise and the power cable to the board got hot. VERY BAD!!!
When the smps is not connected to the board it does not make any sounds... when its connected it makes a very bad sounding sound.
Do you think my SMPS is fried too or is it fine? I'm fairly sure the board is fried but what do you think? Can it be fixed? I've got a friend who's an electrical engineer and has WAY better equipment suited specifically for this kind of stuff and knows his way around a circuit board and diagrams and stuff so, do you think HE would be able to fix the board? Or is it fried and do i need to order a new one....
When the smps is not connected to the board it does not make any sounds... when its connected it makes a very bad sounding sound.
Do you think my SMPS is fried too or is it fine? I'm fairly sure the board is fried but what do you think? Can it be fixed? I've got a friend who's an electrical engineer and has WAY better equipment suited specifically for this kind of stuff and knows his way around a circuit board and diagrams and stuff so, do you think HE would be able to fix the board? Or is it fried and do i need to order a new one....
Sorry to hear about your mishap. I have my board wired up in a small Hammond enclosure (a Hammond 1590D, Aluminum, Diecast). I tried running the board without ventilation, even though it sported a running fan. No way! The enclosure got hot after about an hour. I drilled in a bunch of ventilation holes and it's worked perfectly ever since.
I hope you have success with your next board. I use a Hakko 936 soldering station--it's very simple and totally bulletproof in operation. I used to use Wellers, but I like the Hakko more, especially for the price. You might want to consider spending money on a better-than-decent soldering iron if you plan on doing more of this fun stuff. Good luck!
I really think this is the way to go:
Coilcraft SER2915L-103
The thing is solid and protected and made by people who really know what they are doing.
Coilcraft Inductor Finder
.
Coilcraft SER2915L-103
The thing is solid and protected and made by people who really know what they are doing.
Coilcraft Inductor Finder
.
I tried those Coilcraft 2900
I tried those Coilcraft SER2900 at 10uH. I could go back and try them at the lower value that I am running now but have finally "settled" on the Wurths which sound better and fit on the stock pads. The SER2900 are very well built and rated for huge current and the price is good but had a strange, diffuse blurring in the highs which made me think that maybe the wires in the coil were vibrating in the strong, low turn field. I tried super gluing the coils but couldn't get them to clean up. They aren't terrible but I like the sonics of the Sagami better and was listening to the Wurth's at 3.5uH last night and they are amazing. Getting very close to the depth and transparancy of the air toroids at 6 uH. The zobel becomes important at such low values with the difference between 12,15 and 20 ohms clearly audible. I'm at 15 now and may end up at 14 which if you really wanted to fine tune the amp, might change some with every different speaker. Lower ohms gives a thicker sound and too high a resistance starts to glare.
Is this a high permeability core? I've used high permeability cores and found that they caused distortion in the audio above about 5KHz. Easily seen on the o'scope. That was at a pretty hot level, so it did not show up all the time. Don't know if it was vibration or something else.
Maybe that's what was happening to you.
One of the things I like about the Tripath chips is that they add very little coloration to the sound. Once you get the output coils (or the complete output filter, if you want to go that far) upgraded, you can do the voicing of your amp very simply with the input caps. The amp will faithfully reproduce the sounds the caps feed it.
If you like the warmth, body, and richness of your Dynaco, you might like the Sure better with some good paper-in-oil input caps.
If you like silky smooth with incredible dynamics, you might try teflon caps.
If the Wimas are close, but not quite there, you might want to upgrade to the Dayton poly/foil caps.
Personally, I'm vacillating between the Daytons and some old WestCap paper-in-oil "Vitamin Q types." The Daytons are better, purer, caps. But the PIOs have some magic.
-dr_vega
Hi perm cores
All of these small shielded bobbins from Coilcraft, Toko, Sagami and Wurth are high perm. As high as the 200-300 range? The SER2900 10uH only has 7 turns on a 6mm bobbin. The other weird thing is the flat section wire. I am generally a fan of rectangular wire core shapes but Coilcraft have somehow stretched it around to wind it on edge which makes a big difference in the field between the inner and outer edges of each turn. I would be interested to hear the exact same materials flat wound instead of edge wound. I think the high perm is why the air cores generally sound better. The Ferroxcube cores are 115 perm and still sound really good due to the gapped ferrite construction but are large, have to be hand wound, and leak just as bad as an air core so why use them? The Wurths were recommended on a tip or I would have never have even looked at them but do sound really good. I had some nice custom 12ga. air cores at 8uH made up by Jantzen who is great to work with and reasonably priced but even after unwinding them down, they can't match the sonics of my Belden air torroids and are a toss up with the 3.5uH Wurths so we might as well go with the Wurths since they are shielded and fit on the pads. The ultra low value does allow some 660KHz into the speaker wires but I can still receive a AM 660 station on a hand held radio as close as 2-3 feet from the amps or speaker cables. The transparency and micro dynamics of these modified Sure 2X100 amps are world class. Probably not as liquid in the upper mids as class A or tubes but a no brainer for the money and efficiency, especially for active cross where six or more channels are needed.
Don't forget that on these amplifiers, where the NFB is taken from before the output filter, that the optimum values of L and C in the filter varies with the load impedance. For example, a simple SPICE simulation of the filter shows that for 4R loads, the best inductor value is around 10uH, whereas for 8R loads, a value of around 16uH is better. (Obviously the capacitance values also need also to change to give the optimal load matching and the flattest response). I'm a little surprised that some people seem to think that you can tune the output filter by ear! A simple frequency spectrum sweep into the desired load is better. (There's plenty of s/w out there to do this if you have a good sound card to use as the audio interface. I use ARTA for the measurement s/w.)
Graphs
Unfortunately, I have found that graphs of the flatest frequency response don't tell you what different components will sound like. If only it were that easy. Different makes and models of coils, caps,whatever, with the same value, sound very different.
I am thinking of buying this amp board, but I would like to know a few things first.. Is it possible to power it from a 2x24v 160VA toroid transformer + one or two bridge rectifiers? And probably some big filter caps.
Is it possible to connect the 24v outputs in parallell and get out the whole 160VA? Or can I use only one 24v output and get only 80VA? If I can connect them in parallell, should I connect the AC wires from the both windings together or connect them together after the bridge rectifiers, + to + and - to -?
And a second question: does the board have any "speaker protection" if the circuit fails in some way? Or do I need to build one myself? How should it be built properly? Wouln't like to blow up my new Tannoy F1's...
Is it possible to connect the 24v outputs in parallell and get out the whole 160VA? Or can I use only one 24v output and get only 80VA? If I can connect them in parallell, should I connect the AC wires from the both windings together or connect them together after the bridge rectifiers, + to + and - to -?
And a second question: does the board have any "speaker protection" if the circuit fails in some way? Or do I need to build one myself? How should it be built properly? Wouln't like to blow up my new Tannoy F1's...
Ahhh...a beacon of reason shines through the fog of confusion.
Check out the graph below, which shows actual measurements of a TC2000-TP2050 amp for various loads (2ohm to 8ohm). There is a peak at 6 ohms (red line) that can be noticed with listening tests, yet the response is flat at 7 ohms (cyan line). So obviously, the load impedance is a variable. Since the impedance of most speakers will vary with frequency, the type of high frequency driver being used in the test is a variable. And, of course, the inductance value is a variable. Inductors will have different saturation levels, so listening level is a variable. With this many variables it is difficult to make meaningful comparisons of cores types or inductance values with just listening tests.
Peak
Why does your sim show such a different peak at 6 that doesn't follow the trend of the other loads on either side of it? My simulations always show a continuous trend of increasing peakiness which would show the 6 as a little more rolled off than the 7.
.
You want to talk about a peak, add a minimal amount of inductance, 100uH, to the load as any speaker driver will actually have, and try getting rid of that peak in your filter with any load.
A peaking output filter that is flat at 20K may be necessary for the best sound from a wide variety of speakers and will sound much better than an aggressive textbook filter with no peak which is down at 20KHz with increasing load.
Why does a 7db peak at 60k have to be seen as a bad thing?
Why does your sim show such a different peak at 6 that doesn't follow the trend of the other loads on either side of it? My simulations always show a continuous trend of increasing peakiness which would show the 6 as a little more rolled off than the 7.
.
You want to talk about a peak, add a minimal amount of inductance, 100uH, to the load as any speaker driver will actually have, and try getting rid of that peak in your filter with any load.
A peaking output filter that is flat at 20K may be necessary for the best sound from a wide variety of speakers and will sound much better than an aggressive textbook filter with no peak which is down at 20KHz with increasing load.
Why does a 7db peak at 60k have to be seen as a bad thing?
I might reverse that: with this many variables, the only way to make meaningful comparisons is with listening tests.
I don't play my SACDs to watch an occilloscope. I play them to listen to music. Every system, no matter how good, has distortions caused by dozens (or hundreds) of components and component interactions.
Reducing all distortion to the lowest amount possible does not necessarily produce the best sound. Often the best sound comes from balancing distortions against each other, i.e., offsetting a dip in one component with a resonance in another. That is the main concept behind the Thiele-Small speaker formulae, for example. It's also how we "voice" a system with tube rolling, capacitor rolling, etc.
Measurements can help you move in the right direction, formlae can keep you from melting expensive parts, but, ultimately, you have to listen to the stuff.
The best system is the one that makes you grin.
-dr_vega
It's not a sim. It's a real measurement from a circuit that has basically the same topology as the Tripath data sheet.
Maybe it's not, but my understanding from those who were doing the testing is that it resulted in audible sibilence. Can't verify it myself...
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.