| Russ White |
Here it is folks. Nothing special to look at. I did not want to spend too much time on an enclosure until I knew how it sounded. :) There will be a nicer one to follow.
Well I can safely say I am blown away. This amp sounds far better than my $2,000 Denon.
Here are the specs:
1 400VA toroid trafo with 21V dual secondaries.
2 parraleled LM4780 amp PCBs from Brian.
Both rails are fused at 6 amps fast blow.
The mains are fused with a 3 amp slow blow.
I put a very quite ball bearing 12V fan on the side with a vent on the other. This thing runs nice and cool.
To power the blue LED and fan I wound 18 turns of black hookup wire around the toroid which got me about 5.5V.
This thing just sounds way too good. No hum at all. Gain may be a little high, I may adjust it.
Here are some tips:
1) The LM4780 is not isolated so make sure you use an isolator. I learned tis this hard way (thanks for helping me out Brian!)
2) Brians Kit so rocks. Get one, or a few if you have a surround reciever with pre-outs. :)
3) Prepare to be astounded.
4) Grab a good brew, Antlic Brewing Company's "Coal Porter" makes listening to a new amp even better. :)
 |
|
|
| BrianGT |
Pictures not coming through properly, here are fixed links:
It will be nice to see how the "big cap" kit works out for you once I get the boards.
--
Brian |
|
|
| G4ME |
Looks good, you have more information on the cap upgrade, and where did you score the transformer?
Although, I prefer a nice Shipyard, but alas brewing is my day job to fund my audio adventures. |
|
|
| Russ White |
Retrying images....
|
|
|
| Russ White |
Thanks.
I am going to be build another dual mono amp using Brian's snubberized PS board and the LM4780 amp PCBs. I will compare the two (regular PS vs "Big Cap") side by side.,
For now I see the standard PS version will be stiff competition. It is very detailed, with plenty of bass punch. My wife was just amazed. She couldn't stop saying "That little chip is doing all that?" I had to grin. :D
I have not had the pleasure of trying "Shipyard" but I will certainly try to rectify that! I am not sure I can get it in Nashville though. I hope I can!
The trafo I got from "Antek" who deals on ebay. His trafos are very nice. He even has some really good high voltage trafos for the tube guys.
One other note - I only used one of the rectifier boards for the two amps. The other I used for the third secondary (my own) so I could use the DC fan and the blue LED. |
|
|
| Russ White |
Funny thing. My brother in law was visiting and brought me 6 cases of beer. In amongst them is some "shipyard".
I didn't even make the connection until now.
But I have some "Bluefin" and "Prelude". |
|
|
| Upupa Epops |
| To Rus : What is your age ? |
|
|
| G4ME |
nice i will look into that guy on ebay see what he has for my 3886 when i start the plung.
I was going to say, you can find atlantic brews in TN but no Shiypard? odd, atlantic is a pretty small micro brew up the coast, where shipyard is the third largest in New England. But if you had someone bring them down to you then it makes more sense.
Looks like you are a fan of the Stouts and Porters, so am I, i can't get enough of them. Prelude is personally my favorite, it drinks like a meal and sticks with you. Sadly its a seasonal and not offered year round.
Enjoy the Brews, and let us know how the snubber compeats with the standard PS |
|
|
| Alcaid |
| Antek on ebay is really John Ango who is behind par-metal. I have ordered from him a couple of times. Still waiting for my last order (a 2x21V 400VA toroid and a cabinet for my amp). Waiting for an international shipment really sucks when you can't afford expensive shipping methods..... :( |
|
|
| Russ White |
| Upopa, I am 35. :eek: jeez I am old! :D |
|
|
| Russ White |
| Alcaid, that is exactly the trafo I got. I think the 21V rating is conservative. Its a great trafo. I hope it gets to you soon. I know waiting is a pain. :smash: |
|
|
| Upupa Epops |
| Is this amp your first work in electronic ? |
|
|
| Russ White |
| Yes, it is. I am a complete noob. I still have a lot to learn. I have been reading a lot though. |
|
|
| Upupa Epops |
| So good luck :cool: . But remember that it is long way :cool: . |
|
|
| lgreen |
Where did you get the chip isolators to insulate the chip from heatsink? what kind did you use?
Can't find them anywhere |
|
|
| Vikash |
| Nice one Upupa. Always encouranging. |
|
|
| Russ White |
I made my own. Use simply took out my scissors and cut about 1.5" x 1" square of electrostatic plastic from a bag national semi sent some samples in. I didn't actually measure it, I simply trimmed it until it was just slightly larger than the lm4780. I then cut out two little holes with an hobby kinfe with a sharp blade for the 4-40 machine screws. I smeared some silcon thermal transfer grease on the heatsink(well the 1/4" aluminum sheet) and applied the plastic to that, the grease kinda works like glue :). Then I applied some of grease to the LM4780 and mounted it. Cheap, and works great. :) Thermal transfer seems very good.
One other important note.... Do not use artic silver!!!! It will conduct!! The ceramic version should be OK, but none of the the others. I had some get onto the pins of a LM4780 and completely blow them away when it shorted!!! Bad news.
Personally I would stick to the silicon compound as it is a nice safe choice. |
|
|
| G4ME |
| is it recommend to isolate chips from the heat sink, or did you notice a HF oscillation, and this was a solution? |
|
|
| Vikash |
| you used a cutout from an electrostatic bag to isolate the chip? Will it not melt/degrade quite quickly in time? |
|
|
| Russ White |
the LM4780 package has V- routed to the back of the package, so if you do not isolate you will have negative voltage on you heatsink.
I my initial version I had two independant PSs and this created problems including oscilation. Eventually I accidentally gounded the heatsink with the LM4780s still unisolated and that promptly blew up my two nice PCBs and chips :( :smash: :smash: :bigeyes:
Luckily Brian is a great guy and got me a couple more. :D
The only way I would mount the LM4780 unisolated is if you could ensure that your heatsink was completely isolated.
I am a noob though, so my assesment could be wrong.
I any case, my amp has absolutely no humm right now, and when it was working unisolated it did have a slight humm, and was more susceptable to RF noise and such. |
|
|
| Russ White |
The plastic I used is pretty thick, and very duable, it does not seem very prone to heat damage, and the chips/interface have not gotten over 70C measured with my IR thermometer. The plastic should serve just fine.
I cannot take credit for this trick. It was a national semi engineer who suggested it, saying he had used the technique for quite a while. I had called nat semi to try to find out where I could get mica isolator for the 27 pin package. I have yet to find a source...
In any case I will watch it over time and report the results.. |
|
|
| jackinnj |
| quote: | Originally posted by lgreen
Where did you get the chip isolators to insulate the chip from heatsink? what kind did you use?
Can't find them anywhere |
Digikey carries "Bergquist" rectangular insulators -- the BER136 is just a little too big for an LM4780. you will have to punch some holes (or cut out with your wife's needle-point scissors :) -- she will probably stab you with them if she finds out they are in the lab.
the C/W of the SilPad 400 is about 0.45 so take this into consideration when sizing your heatsink.
btw, the Bergquist BER107 is ideally sized for the LM3886 and LM3875.
http://dkc3.digikey.com/PDF/T051/0640.pdf |
|
|
| Vikash |
| quote: | Originally posted by Russ White
The plastic I used is pretty thick, and very duable, it does not seem very prone to heat damage, and the chips/interface have not gotten over 70C measured with my IR thermometer. The plastic should serve just fine.
I cannot take credit for this trick. It was a national semi engineer who suggested it, saying he had used the technique for quite a while. |
Top tip. :up: |
|
|
| BrianGT |
| quote: | Originally posted by jackinnj
Digikey carries "Bergquist" rectangular insulators -- the BER136 is just a little too big for an LM4780. you will have to punch some holes (or cut out with your wife's needle-point scissors :) -- she will probably stab you with them if she finds out they are in the lab.
the C/W of the SilPad 400 is about 0.45 so take this into consideration when sizing your heatsink.
btw, the Bergquist BER107 is ideally sized for the LM3886 and LM3875.
http://dkc3.digikey.com/PDF/T051/0640.pdf |
I had looked at this earlier (the BER136), but saw that it was an electrically conducting material:
| quote: | | Q-Pad 3® is a thermally conductive, fiberglass reinforced graphite imbedded elastomer which conforms to surface textures extremely well. This material eliminates problems associated with thermal grease such as contamination of electronic assemblies and reflow solder baths. Q-Pad 3 may be installed prior to soldering and cleaning. (Electrically Conducting Material) |
Wouldn't this cause problems?
--
Brian |
|
|
| Russ White |
| quote: | Originally posted by BrianGT
I had looked at this earlier (the BER136), but saw that it was an electrically conducting material:
Wouldn't this cause problems?
--
Brian |
Based on my sad experience I would be wary of it.
I would certainly prefer Mica or something else noncunductive.
What I wonder is if there is ever a reason you would want the negative voltage from the chip to the heatsink. Just seems there must be a reason they designed it that way.
I wonder if you could connect you V- to the heasink instead of to a PCB or something? Just not sure how practical that is.
Its a puzzlement. :confused: |
|
|
| ransom peek |
Wouldn't a transformer at 24 to 25 vac offer increased upside power over the 21 volt transformer used here, and thus be prreferred if all else (power, price, etc) was the same?
Is a 330 va transformer adequate for a stereo version of the 4780?
Does Brians LM 4780 stereo kit come with the required insulator for the 4780? |
|
|
| jackinnj |
the BER136 has a resistance of about 400 ohms -- so you aren't going to get much protection -- you're right.
the BER107 doesn't seem to conduct. |
|
|
| digi01 |
| quote: | Originally posted by Russ White
Yes, it is. I am a complete noob. I still have a lot to learn. I have been reading a lot though. |
wow,good start!congratulations:cool:
ZANG |
|
|
| Russ White |
| quote: | Originally posted by ransom peek
Wouldn't a transformer at 24 to 25 vac offer increased upside power over the 21 volt transformer used here, and thus be prreferred if all else (power, price, etc) was the same?
Is a 330 va transformer adequate for a stereo version of the 4780?
Does Brians LM 4780 stereo kit come with the required insulator for the 4780? |
Yes, 24-25 volt secondaries would allow for higher power output. As it is this thing seems very potent with the 21V secs. :)
I would think the 330VA would be enough, but I will let others answer that.
My kit did not include any isolator, but I am not sure if they normally do or not as I asked for a kit with no LM4780 chip since I have a few in stock. |
|
|
| steenoe |
Congrats, with your new amp. Nice looking. BUT:| quote: | | cut about 1.5" x 1" square of electrostatic plastic |
I am not sure this will work in the long run, especially when looking at the heatsinking. The 4780 are capable of quite a few amps of power and need proper heatsinking. I think you should get some pads that is designed for heat transfer! Both Silicone and Capton are available as sheets. Cutting your own washers from that kind of stuff would be more appropriate. I feel like quoting Janneman on this:| quote: | | This is a Politically Very Uncorrect post. |
If you intend to run this amp at full throttle, I believe you will discover the protection circuit, of these chips:)
All the best
Steen. |
|
|
| steenoe |
Forgot: If you cant get your hands on the right stuff for washers,
just let me know. I can send you a strip of Capton:)
That would make me sleep tight to night!
Steen.:) |
|
|
| Russ White |
| quote: | Originally posted by steenoe
Congrats, with your new amp. Nice looking. BUT:
I am not sure this will work in the long run, especially when looking at the heatsinking. The 4780 are capable of quite a few amps of power and need proper heatsinking. I think you should get some pads that is designed for heat transfer! Both Silicone and Capton are available as sheets. Cutting your own washers from that kind of stuff would be more appropriate. I feel like quoting Janneman on this:
If you intend to run this amp at full throttle, I believe you will discover the protection circuit, of these chips:)
All the best
Steen. |
I appreciate the advice, and I have taken it to heart.
I have already ordered a couple of sheets of silicon pads from digikey.
But... for what its worth... I have run the amp at near full throttle for nearly an hour out in the garage and as mentioned earlier the comination of the large aluminum plate and the fan kept the thing nice and cool. The plastic actually seams to conduct heat pretty well. The thermometer registered ~70C at full tilt.
I plan on observing it for a few days.
I will probably stress test it again soon. |
|
|
| jackinnj |
| quote: | Originally posted by steenoe
Congrats, with your new amp. Nice looking. BUT:
I am not sure this will work in the long run, especially when looking at the heatsinking. The 4780 are capable of quite a few amps of power and need proper heatsinking. I think you should get some pads that is designed for heat transfer! Both Silicone and Capton are available as sheets. Cutting your own washers from that kind of stuff would be more appropriate. I feel like quoting Janneman on this:
If you intend to run this amp at full throttle, I believe you will discover the protection circuit, of these chips:)
All the best
Steen. |
Yes, the PDMax equations on the National Semiconductor website will indicate the power which the chip should dissipate apart from the load -- if they are run flat out they require a lot of heatsinking.
and I have pointed out that there is some latency period in the SPIKE protection where it is thinking of kicking in but can't quite make up its mind -- and the wave shape gets niced and fuzzed up -- |
|
|
| steenoe |
| quote: | | The thermometer registered ~70C at full tilt. |
That is quite a bit too much, I think. The normal recomandation is to not go over 60c. But Diy'ers are a frightless breed, aren't they?
70C will shorten the lifespan of the chip, but having the protection circuit working starts me wondering:xeye: How and where did you measure that temp? The 60C temp is for a class A amp, and a chipamp like this, should not reach so high temps!
All the best
Steen. |
|
|
| BrianGT |
| quote: | Originally posted by ransom peek
Does Brians LM 4780 stereo kit come with the required insulator for the 4780? |
I haven't found any recently that would work well with the LM4780 kits, so I haven't been including insulators with all kits.
| quote: | Originally posted by jackinnj
the BER136 has a resistance of about 400 ohms -- so you aren't going to get much protection -- you're right.
the BER107 doesn't seem to conduct. |
The BER107 doesn't seem to fit right for the LM4780, and there is a hole in the middle.
I e-mailed most major companies that offer heatsink pads, and National, and could not find a good solution for the LM4780, except for buying sheets and cutting them in pieces.
--
Brian |
|
|
| Russ White |
That was the temp directly behind the chip measured with an IR probe.
The aluminum sheet is 1/4" thick 5" high and 15" long. The average temperture after the hour stress test was < 50C which seems resonable, is it not? |
|
|
| steenoe |
| quote: | | except for buying sheets and cutting them in pieces. |
That is excactly, what you need to do, with an abnormal chip like this.
Steen. |
|
|
| Russ White |
| Just checking things out I just popped the top off and felt the heatsink, been listening to the amp all day at normal to slightly high levels and the heatsink is only slightly warm as I feel it directly behind the chip. So I think it should be OK. Really it is running cooler than I expected based on stories I have read. |
|
|
| steenoe |
| quote: | | < 50C which seems resonable, is it not? |
Yes, that is ok! If your heatsink average about 50C at full throttle
for an hour, you might be safe after all;) Just take care that the fan doesn't stop, though:)
All the best
Steen. |
|
|
| jackinnj |
| quote: | Originally posted by BrianGT
I haven't found any recently that would work well with the LM4780 kits, so I haven't been including insulators with all kits.
The BER107 doesn't seem to fit right for the LM4780, and there is a hole in the middle.
I e-mailed most major companies that offer heatsink pads, and National, and could not find a good solution for the LM4780, except for buying sheets and cutting them in pieces.
--
Brian |
I think that what I wrote was that the BER107 will work with the LM3886 and LM3875.
There are smaller silpads which look as if they are made for a TO-220 regulator. If you butt two of them together they will handle an LM4780.
Below is the BER107 |
|
|
| steenoe |
| quote: | | If you butt two of them together they will handle an LM4780. |
That is a possibility, personally I would hessitate to go that way. If there is a gap or worse an overlap, that would not be a great solution. Also, silicon pads are not the best "heattransferrers". A thin mica washer is better. I feel sure the plastic bag thing, is even worse though. It is for sure not uniform enough in the thickness! Go for the real thing.
Good luck
Steen. |
|
|
| kestrel200 |
| quote: | Originally posted by steenoe
That is a possibility, personally I would hessitate to go that way. If there is a gap or worse an overlap, that would not be a great solution. Also, silicon pads are not the best "heattransferrers". A thin mica washer is better. I feel sure the plastic bag thing, is even worse though. It is for sure not uniform enough in the thickness! Go for the real thing.
Good luck
Steen. |
So... Where can I find/buy mica washers? Or for that matter even silicon that can be cut to size???
:xeye: |
|
|
| BrianGT |
| quote: | Originally posted by jackinnj
I think that what I wrote was that the BER107 will work with the LM3886 and LM3875.
There are smaller silpads which look as if they are made for a TO-220 regulator. If you butt two of them together they will handle an LM4780.
Below is the BER107 |
Sorry I misunderstood your post earlier, I guess I didn't read it clearly. The BER107 looks perfect for the LM3875/LM3886.
I hope something else comes out soon for the LM4780, as I don't know if I would trust 2 butted together of the BER107.
--
Brian |
|
|
| kestrel200 |
| quote: | Originally posted by BrianGT
Sorry I misunderstood your post earlier, I guess I didn't read it clearly. The BER107 looks perfect for the LM3875/LM3886.
I hope something else comes out soon for the LM4780, as I don't know if I would trust 2 butted together of the BER107.
--
Brian |
So Brian can you answer my question? Where can we get silicone that we can cut to size since it's not included in the kits? |
|
|
| mhb |
The Bergquist BER118-ND is my fav for the 3876 and 3886. It's made of viton which is superior to the 107 silicone/fiberglass.
Digikey has 'em, here is the link:
LM3876/3886 Pad
Bergquist does sell viton by the sheet but not in DIY quantities (not that I know of, anyway).
Cheers,
m |
|
|
| kestrel200 |
so.. to build your kit I have to buy this from Digikey at:
Price Break
2 Unit Price
188.72000 Price
377.44
:bigeyes:
Does anyone have any other ideas?? If not, I think I'm going to have to cancel my order. |
|
|
| BrianGT |
| quote: | Originally posted by kestrel200
so.. to build your kit I have to buy this from Digikey at:
Price Break
2 Unit Price
188.72000 Price
377.44
:bigeyes:
Does anyone have any other ideas?? If not, I think I'm going to have to cancel my order. |
I accidently removed the link, here is the digikey page:
http://dkc3.digikey.com/PDF/T051/0641.pdf
starts at $9.98 for a 4"x4" piece.
BER161-ND or BER162-ND
The CPU pad I was looking at earlier was the one on the top of the page, but it just looks like thermal double sided tape with not too good thermal properties compared to pads.
--
Brian |
|
|
| kestrel200 |
| quote: | Originally posted by BrianGT
I accidently removed the link, here is the digikey page:
http://dkc3.digikey.com/PDF/T051/0641.pdf
starts at $9.98 for a 4"x4" piece.
BER161-ND or BER162-ND
The CPU pad I was looking at earlier was the one on the top of the page, but it just looks like thermal double sided tape with not too good thermal properties compared to pads.
--
Brian |
Now your talking!!!
;)
So I only need one pad. Right?
I can't wait to get the kit now! I bought & built two of your 1st generation 3875 Premium kits and they're great. I have a 600VA 25V Transformer allready waiting for this baby.
THX! |
|
|
| BrianGT |
| quote: | Originally posted by kestrel200
Now your talking!!!
;)
So I only need one pad. Right?
I can't wait to get the kit now! I bought & built two of your 1st generation 3875 Premium kits and they're great. I have a 600VA 25V Transformer allready waiting for this baby.
THX! |
One of the 4"x4" sheets should work for quite a few channels.
--
Brian |
|
|
| Russ White |
I only used the electrostatic thick plastic as a quick fix and only for a temporary solution while I wait for my isolator sheet from digikey.
Still, I have to say it is performing quite well, This amp have 20 or 30 hours of 4ohm load on it, and it is holding very well indeed. Thermal transfer is very good to the heat sink.
If you ever needed a solution to hold you over until you got some isolators it works quite well. :cool:
It would be intersting to see a scientific comparison between my lowly plastic bag interface/isolator and the expensive stuff. You might be surprised. :D |
|
|
| Russ White |
| When I leave the amp on, but turn off the Cd player after about 1 second or two I get a momentary humm, and then my positive rail fuse blows.... hmmmm... Should I be grounding my inputs somehow? Maybe a cap to chasis ground from input ground? Now I am a little confused... :confused: |
|
|
| Peter Daniel |
This may depend what your CD player does to the output when it's switching off. You might check DC offset first.
What type of volume control to the amp are you using? |
|
|
| Russ White |
Hi Peter,
In this case I am using the headphone amp which is integrated with the cd player as the pr-amp directly to my GC. |
|
|
| Russ White |
| DC offset just measured is 52mv left channel 47mv right, |
|
|
| Peter Daniel |
In that case the input of the amp is at mercy of your CD player output. If for instance, the output is completely disconnected (by CDP muting circuit), you will be getting hum (in your amps).
How is the offset, when you switch off the player, does it change? |
|
|
| Russ White |
| OK, I popped a 8 amp fuse in positive and negative rails and measure offset while turning off CDP. It did just jump off the scale.... and then even the 8 amp fuse blew. Looks like I need to use a real preamp, or always turn off the GC first. :xeye: |
|
|
| Peter Daniel |
| You might try install a coupling cap at the amp's input and see if this helps. A value of 2uF or more woud be recommended. |
|
|
| Russ White |
| quote: | | You might try install a coupling cap at the amp's input and see if this helps. A value of 2uF or more woud be recommended. |
Thanks Peter.
OK, I had a couple of non-polar 2.2uf electrolitics in my box. I looked at the LM4780 spec sheet to see how they had their input couplng cap. It looks like it is just connected in series right at the input. So that is what I did. I soldered one end directly to the jack and the other to the hookup wire. I then just taped it up the help prevent shorts.
Well, now the humm when I turn off the CDP is vastly attenuated to where it is not a problem, and things sound good, but....
The DC offset is now measuring 150mv on both channels.
Just to double check I removed the coupling cap and it went back down to 30-50mv where it was before.
Did I do something wrong?
 |
|
|
| Peter Daniel |
| quote: | Originally posted by Russ White
Did I do something wrong?
|
You did nothing wrong, but now the output from your CDP is separated from amp input by a coupling cap. So the impedance the amp see is 22k and with that the offset is about 150mV, quite normal for a parallel LM4780 without a cap from - IN to ground.
When you connect your CDP directly, the output impedance of the player in parallel with 22k resistor at the amp's input , produces much lower combined impedance and this makes for lower DC offset, as it's been discussed here: http://www.diyaudio.com/forums/show...1431#post571431
The LM4780 kit was developed mainly for power amp only implementation, so the combined impedance of source and amp should be always less than 22k (at amp's input) and DC offset should not be a concern.
In your particular setup, you might try to reduce 22k resistor to the value that generates DC offset less than 100mV and you should be fine. |
|
|
| Russ White |
Ahhh!! Thanks for the pointer. I was really worried I connected it up wrong. Though I am sure it would be tough to mess it up, one should never underestimate my messing-up abilities. :bigeyes: :D
| quote: | | quite normal for a parallel LM4780 without a cap from - IN to ground. |
So if I wired another 2.2uf cap from the -IN pin to the CHG pad on the PCB would that help?
Or, if you think its better just to change that 22K resistor...
I am still a rank ohm's law amature, so can you guide me a little on the formula to find the right value to replace it with? I am not sure how to calculate the impedance I introduced with the 2.2uf cap. :xeye: |
|
|
| Peter Daniel |
As you see here http://www.national.com/ds/LM/LM4780.pdf National recommends 47u for that cap (between 680R resistor and ground in 4780 kit) and you probably shoudn't go with less than 22u.
If the 10k resistor at amps input would be sufficient to lower the offset it should be fine that way. You might try to use some 47u electrolytic as to Nationals recommendation and this should work OK too.
The 4780 kit was made with minimalistic approach in mind, as the ommiting the cap produces better sound. |
|
|
| Russ White |
Ok it was right in front of my face. Stupid me. :smash:
Ahhh yes!!! I had seen that cap and wondered its purpose, now it is much more clear!!!
Ok, so with this PCB Is it possible to P2P wire the 47uf shunt capicitors if I wanted to experiment? Would I have to cut a trace for each cap?
I will try the 10K resistors for now. |
|
|
| BrianGT |
| quote: | Originally posted by Russ White
Ahhh!! Thanks for the pointer. I was really worried I connected it up wrong. Though I am sure it would be tough to mess it up, one should never underestimate my messing-up abilities. :bigeyes: :D
So if I wired another 2.2uf cap from the -IN pin to the CHG pad on the PCB would that help?
Or, if you think its better just to change that 22K resistor...
I am still a rank ohm's law amature, so can you guide me a little on the formula to find the right value to replace it with? I am not sure how to calculate the impedance I introduced with the 2.2uf cap. :xeye: |
With regards to the input capacitor, this create a simple high pass filter, which is set by a combination of the input impedence and the capacitor value. Use f= 1/(2*pi*r*c) for calculating the freq.
having too small of a cap will roll off the bass too much. I would choose 4.7uF or 10uF.
Alternately, the optional cap in the feedback network (notated as Ci in the datasheets) creates a high pass filter using the same equation. I added an optional placement for this component on the new LM3886 pcb set. For wiring the other boards P2P, you will stick the cap in one of the resistor leads, and the resistor in the other one, and then put the r and c in series.
--
Brian |
|
|
| Russ White |
Brian/Peter thanks you guys. You have have educated me today. I Will audition both mods and see which I prefer. Is a 35V 47uf OK for the shunt capacitor? Or should I get 50V caps?
Just to be sure I havent mucked it all up in my head here are the two mods:
#1 The easy way out
A) replace both resistor at R1 and R2 with 10K 1/4 watt resitors.
#2 The national/shunt way
A) take out R3 and R5
B) Take a 680R resistor and solder one lead only to the pad toward to the chip.
C) Solder the other lead of each 680R resistor to a 47uf electrolitic at positive pole.
D) Solder the negaitive pole of the 47uf cap to the remaining pad for R3 and R5.
According to my calculations the corner frequency for the highpass filter at imput with the 2.2uf cap I used is F=3.2734459706272174 Is that too high? |
|
|
| BrianGT |
| quote: | Originally posted by Russ White
Brian/Peter thanks you guys. You have have educated me today. I Will audition both mods and see which I prefer. Is a 35V 47uf OK for the shunt capacitor? Or should I get 50V caps?
|
Since your power rails are below 35v, the 35v cap would work for testing, but if it was my amp, I would prefer a 50v cap for long term reliability.
| quote: | Originally posted by Russ White
#1 The easy way out
A) replace both resistor at R1 and R2 with 10K 1/4 watt resitors.
|
R1 and R2 are the feedback resistors, and should not be removed. R4 is the resistor to ground on the input. Here is the schematic:
http://www.chipamp.com/lm4780.pdf
You will want to try replacing R4 with 10k.
| quote: | Originally posted by Russ White
#2 The national/shunt way
A) take out R3 and R5
B) Take a 680R resistor and solder one lead only to the pad toward to the chip.
C) Solder the other lead of each 680R resistor to a 47uf electrolitic at positive pole.
D) Solder the negaitive pole of the 47uf cap to the remaining pad for R3 and R5.
|
For this, you can just pull up 1 end of each 681 ohm resistor (R3 and R5 as you mentioned), and connect the removed end to the electrolytic cap, and put the other lead into the resistor hole on the board. You will want to observe the polarity, and put the + lead to GND, as mentioned in the application note. This means that you should pull the resistor leads out of bottom holes (closest the the GND pad) and put the + cap leads in there.
| quote: | Originally posted by Russ White
According to my calculations the corner frequency for the highpass filter at imput with the 2.2uf cap I used is F=3.2734459706272174 Is that too high? |
This should be fine, but 4.7uF might be better.
--
Brian |
|
|
| Russ White |
| quote: | | You will want to observe the polarity, and put the + lead to GND, as mentioned in the application note. This means that you should pull the resistor leads out of bottom holes (closest the the GND pad) and put the + cap leads in there. |
Hmmm.... Looking at the parallel application (on page 6) on the datasheet it looks like the electrolitics should have the positive side connected to the resistor... And the negative toward ground.... Am I reading the wrong schematic? Or do I not know how to tell the polarity of the cap?
Thanks for the correction on R4, see I told you I could muck it up. :rolleyes: |
|
|
| Russ White |
Brian - Thanks.
I replaced the r4 resistor with a 10K resistor, but it only dropped the dc offset by about 30mv to 120mv :(
Guess I will try the shunt approach. |
|
|
| Kevin Haskins |
I've found these perform best when you really make efforts to keep them cool. The National thermal protection seems to start working significantly before it actually shuts the chip down due to thermal overload. Making the extra efforts to go overkill on the heatsink and the thermal interface are what I consider a smart move.
I'm stocking these which work fine with even a pair of chips.
http://www.thermaflo.com/bin/exdata...in&ExLength=2.8
I could sell them for about $15 each for those looking for a passive heatsink.
For the thermal interface I looked at the Arctic Silver stuff, both the silver & ceramic based compounds. Contrary to their claims they are electrically conductive with the LM4780 application (but they are awesome thermal conductors).
I've been using the following which I bought in a 12"x 12" sheet for $85.
http://www.thermaflo.com/phase/pdf/8040900web.pdf
It comes out to about $2/chip to insulate but this has about the best thermal performance I've found in an electrically insulating pad. Having the heatsink tied to the V- has proven to be a bad choice even if you isolate the heatsink from the chassis. It picks up RFI & other noise.... |
|
|
| pinkmouse |
| I have gone for the big heatsink approach as well. |
|
|
| kestrel200 |
Ok. I'm about to start on my 3rd Brain GT amp and my first 4780. Should I use Brain's center tap diagram to hook this transformer up with just on rectifier board or is there a way to hook it up using both boards? Or????
Thx |
|
|
| jackinnj |
| quote: | Originally posted by pinkmouse
I have gone for the big heatsink approach as well. |
well, you had better use a big heatsink -- these things cook! Here's a drawing of the bar I made to affix mine to the heatsink:
 |
|
|
| pinkmouse |
I just finished building mine half an hour ago. Fresh tomorrow, I'll check everything then fire them up and see how they get on. I'm greatly looking forward to it. :)
Long term, they will end up driving my B139 TL sub project with active crossovers, but I want to give them a good audition with my main system first. |
|
|
| SCD |
Hello folks:
I have been struggling my way through a kit from Brian and now seem to have a new problem to solve. I previously screwed up my power supply, luckily I got that figured out. Now I have a new challenge. I get a lower frequency oscilation that starts out low then picks up speed. I turn it off before it gets going too fast. I am not sure how to describe it other than motor boating.
Got any ideas. Did I fry something with my power screw up?
Any trouble shooting help is welcome
thanks for the help |
|
|
| Kevin Haskins |
| quote: | | well, you had better use a big heatsink -- these things cook! Here's a drawing of the bar I made to affix mine to the heatsink: |
Nice.... mine doesn't get hot at all. The front of the chip isn't even warm after days of use. I'll try thermal testing into a low impedance resistive load and see how hot I can get the HS. |
|
|
| pinkmouse |
| Check all your soldering, then try again with the input of the amp shorted to ground. Minimised GCs do have a habit of oscillating if the inputs are floating. |
|
|
| jackinnj |
| quote: | Originally posted by SCD
Hello folks:
I have been struggling my way through a kit from Brian and now seem to have a new problem to solve. I previously screwed up my power supply, luckily I got that figured out. Now I have a new challenge. I get a lower frequency oscilation that starts out low then picks up speed. I turn it off before it gets going too fast. I am not sure how to describe it other than motor boating.
Got any ideas. Did I fry something with my power screw up?
Any trouble shooting help is welcome
thanks for the help |
that's what the 10uF/50V electrolytics are for -- National Semi specifically cites "motorboating" as a problem. |
|
|
| jackinnj |
| quote: | Originally posted by Kevin Haskins
Nice.... mine doesn't get hot at all. The front of the chip isn't even warm after days of use. I'll try thermal testing into a low impedance resistive load and see how hot I can get the HS. |
I shouldn't have jumped in prematurely -- my board is using the bridged configuration and this dissipates 4 times the amount of energy as heat compared to a dual-mono board. My apologies. |
|
|
| Kevin Haskins |
| quote: | | I shouldn't have jumped in prematurely -- my board is using the bridged configuration and this dissipates 4 times the amount of energy as heat compared to a dual-mono board. My apologies. |
Oh...yes. That will do it. I'm using the 10uF + 0.1uF cap on the chip per National's tech notes so I've not experienced any motorboating. I never had any problem with it on the LM3875 ala Brian's boards either but did't want to take any chances with my board. |
|
|
| SCD |
Thanks for the comments guys:
I will double check my soldering again, and ground my input. If that does not do it I think I may have fried the capacitors on the power supply board? I suspect it might be a floating ground as I have just mickey moused it together.
Thanks for the help
It is tough being new at this hobby, so many pitfalls to get caught in. |
|
|
| D_GR8_1 |
I got a couple of questions regarding the design:
What is the output power of each channel?
What size transformer is required? Each LM3886 needing about 80VA = 320VA + losses, which means a 500VA should be enough?
Looking at Brian's schematic he has omitted the input cap. Why is that? |
|
|
| Kevin Haskins |
I would think 350-400VA would probably suffice but 500VA is fine.
Power output is going to depend on load & power supply voltage. Look at the National tech notes for more details as they have plenty of diagrams showing power ratings @ a given distortion.
I'll leave it to Brian to explain his design but I didn't use them on my board because they take up a lot of space and I may or may not need DC blocking depending on what is used to feed the amplifier. If you need DC blocking put a couple 1.0uF caps in series with the RCA inputs and you have the problem solved. |
|
|
| Peter Daniel |
| quote: | Originally posted by D_GR8_1
Looking at Brian's schematic he has omitted the input cap. Why is that? |
For a same reason that there is no caps on LM3875 board: better sound quality. If someone is uncertain about DC offsets of the sources, coupling caps can be always added outboard.
Page 13 of the datasheet shows available power levels depending on supply voltage: http://www.national.com/ds/LM/LM4780.pdf
With this board, the chips are connected in parallel. |
|
|
| D_GR8_1 |
Peter:
is it possible to get the PCB pdf of the LM4780 board? |
|
|
| D_GR8_1 |
Thanks.
I can do this on a double sided PCB right?
I mean, I dont need any through hole connections or whatever. |
|
|
| D_GR8_1 |
Peter, comparing the pdf of LM4780 on
Chip Amp
and the PCB layout you just sent me, I can see some disrepancies.
It is missing the Zobel network as well as the muting resistor.
The pdf you sent me, is that your design or Brian's?
Also, what are the differences in designs? |
|
|
| Peter Daniel |
| quote: | Originally posted by D_GR8_1
The pdf you sent me, is that your design or Brian's?
|
That pdf was posted by Brian here: http://www.diyaudio.com/forums/show...6885#post386885
It is one of the earlier versions of the board. Later it's been modified to include Zobel and nicer trace lines.
I have the latest layout in my files, but it's not converted to pdf yet. |
|
|
| D_GR8_1 |
Once you convert it to pdf, could you post it on this thread please?
Instead of doing it myself and then opening another thread to get people's opinions on it, I would rather print it and etch the pcb. |
|
|
| kestrel200 |
| quote: | Originally posted by kestrel200
Ok. I'm about to start on my 3rd Brian GT amp and my first 4780. Should I use Brain's center tap diagram to hook this transformer up with just on rectifier board or is there a way to hook it up using both boards? Or????
Thx |
Would someone pls help me with this?
Thx |
|
|
| Russ White |
| quote: | | Would someone pls help me with this? |
Use both secondaries on one rectifier PCB to get your negative and positive rails.
Since you only have one trafo you should probably just use one of the rectifiers. This will work fine for both amp PCBs for a stereo setup. |
|
|
| Upupa Epops |
| 3886 is 60 W amp. Typical efficiency of SS amp is about 65 % = 100 W transformer for each channel is enough. 500 W transformer is " overkill size ", which bring only little improving. Gentlenen, stop to be crazy :cool: . |
|
|
| kestrel200 |
| quote: | Originally posted by Upupa Epops
3886 is 60 W amp. Typical efficiency of SS amp is about 65 % = 100 W transformer for each channel is enough. 500 W transformer is " overkill size ", which bring only little improving. Gentlenen, stop to be crazy :cool: . |
Not when you find one at a local wholesale store going out of business for $20! That's less than what I'd have to pay for a 100VA. :D |
|
|
| Upupa Epops |
| If you have at home a crane, go on ! :D ( Sometimes I have feeling, that all electronic DIY hobby make only poor peopple, mainly if they are from USA :D ) |
|
|
| jackinnj |
| quote: | Originally posted by Upupa Epops
3886 is 60 W amp. Typical efficiency of SS amp is about 65 % = 100 W transformer for each channel is enough. 500 W transformer is " overkill size ", which bring only little improving. Gentlenen, stop to be crazy :cool: . |
Put an a.c. ammeter on your GC power supply, load the input with 1 volt of bandwidth limited Gaussian noise, and examine the VA it actually consumes at rated output. A good portion of the power is dissipated by the chip as heat, not music. |
|
|
| Upupa Epops |
| I don't understand - can you say concrete numbers ? |
|
|
| carlosfm |
Upupa, 100VA is small, even for one chip.
We are talking 2.27A (for a 2x22V trafo).
I don't say it doesn't work, but the amp will not be able to drive most speakers, and you must forget 4 ohms speakers completely.
You should dimention a PSU for a 4 ohm load.
Because most of the 8 ohm speakers today have impedance dips to 4 ohms, and some even lower.
Besides bass-refex enclosures, 4 ohm woofers with 8 ohm tweeters is becoming very usual, and then they call this 6 ohms.:dodgy:
Once again, if we are doing our amps, why make the same compromises most brands make on commercial products?
I would recommend 180VA per chip. |
|
|
| Upupa Epops |
| Have your amp efficiency 30 % ? Is it class A ? My 4780 give in parallel mode 120 W RMS into 2 Ohm load with 200 VA transformer, or 2*60 W RMS into 4 Ohm ( 80 W when is driven only one channel ). It is only about physical laws, not about " commerciality " of products. |
|
|
| carlosfm |
| quote: | Originally posted by Upupa Epops
Have your amp efficiency 30 % ? Is it class A ? My 4780 give in parallel mode 120 W RMS into 2 Ohm load with 200 VA transformer, or 2*60 W RMS into 4 Ohm ( 80 W when is driven only one channel ). |
Something's wrong with these "physical laws"...:confused:
How can, even paralleled, the LM4780 double the power from 4 to 2 ohms?
Also, what voltage are you using? |
|
|
|
