LM3875 dual mono kit - a few questions.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I have bought 2 dual mono 3875 kits from madaboutsound.com to make a 4ch amp.

I finished the first 2 channels yesterday but I'm getting 78mv at one set of terminals and 144mv at the other set.

I have triple checked everything and both channels are identical - any idea what I may have done wrong?

I have read that I'm supposed to ground the PSU board but can't see where to ground it so only the amp boards are grounded to the chassis via cg.

Now... This morning I put the other kit together for the other 2 channels, I switched the amp on and smoke started pouring off one of the rectifier diodes... Not good...

Again, I triple checked everything and it's all put together properly. I haven't ever built anything like this before so I have been working very carefully to make sure I do everything properly and don't make any mistakes.

The transformer I'm using is a 500 VA with two 25-0 secondaries, the PSU boards are connected AC1 to +25, AC1- to 0, AC2 to +25 and AC2- to 0.

What would cause the diode to smoke?

On the bright side, I listened to the first amp for a few hours first thing this morning ( I have spare drivers for that particular pair of speakers ;)) and it sounds exceptional, I'm well chuffed. :)
 
This is not unusual values.

According to datasheet the output offset with low ohmish resistors in the feedback and low DC resistance at the input, it can be 1-10 mV times the gain which is 30, 30-300 mV that is. If you them have non optimal resistor values the offset can be even higher. (therefore do I use DC-servo which gives me 0.07 mV(!) offset, every time!)
 
quickshift said:
I have bought 2 dual mono 3875 kits from madaboutsound.com to make a 4ch amp.

I finished the first 2 channels yesterday but I'm getting 78mv at one set of terminals and 144mv at the other set.

Are those measurements without source connected, and if connected, are you sure that source does not generate any DC offset?

BTW, 144mV is very unusual value. Out of 300 chips I measured, I incountered such high offset only 3 times. Chips producing more than 100mV offset are usually less than 10%.
 
Peter Daniel said:


Are those measurements without source connected, and if connected, are you sure that source does not generate any DC offset?

BTW, 144mV is very unusual value. Out of 300 chips I measured, I incountered such high offset only 3 times. Chips producing more than 100mV offset are usually less than 10%.

The measurements are with no source connected. I have also checked the output from the PSU and it's +/- 35.1V on both boards.

Should I be worried that they values are so different?

Thanks peranders

<<QS runs off to look up what a DC-servo is... >>

I think this was with a previous revision of the board but Db Dungeon says 28mV DC offset is about right. As long as it's not going to cause a problem that's OK.
 
You have to also consider, that depending what is connected to the input, those values of the offset may decrease, as offset depends on input shunt resistance as well.

So, if it's active preamp, with no coupling cap, the offset may drop to 20mV or so. If you use a pot at the input, the offset will vary with rotation (of the pot). I find that anything less than 70 mV (or even 100mV) is fine for the speakers, but personally prefer if the offset is less than 50mV. If the offset is excessively high, the chip should be replaced.
 
If you short circuit the input you will see how much voltage offset you have got.If you measure the output and divide it with the gain (32 in your case) you will see if the value is below 10 mV which is technically OK but I presume you will get 3.6 mV.

If you then want to get it lower you may in sert a pretty big capacitor in series with the 680 ohms but this is not a good idea. You can't lower it further without adding any trimming network. Maybe the best solution is to change the LM3875 but then you may mess up the board in the process.
 
Peter Daniel said:
You have to also consider, that depending what is connected to the input, those values of the offset may decrease, as offset depends on input shunt resistance as well.

So, if it's active preamp, with no coupling cap, the offset may drop to 20mV or so. If you use a pot at the input, the offset will vary with rotation (of the pot). I find that anything less than 70 mV (or even 100mV) is fine for the speakers, but personally prefer if the offset is less than 50mV. If the offset is excessively high, the chip should be replaced.

Thank you.

I have measured it with the preamp connected, it's active but I have no idea if it has a coupling cap. Sorry. So much to learn... :)

I haven't used a pot in the amp.

It's now reading 55mV and 19mV which is more sensible. I'll call the shop tomorrow and have a chat with them about it.

I still can't believe how good £35 worth of parts sounds, I can't wait until it's run in and all 4 channels are working. :)

These go pretty close to the top of my "things I wish I had found out about some time ago" list. :D
 
peranders said:
If you short circuit the input you will see how much voltage offset you have got.If you measure the output and divide it with the gain (32 in your case) you will see if the value is below 10 mV which is technically OK but I presume you will get 3.6 mV.

If you then want to get it lower you may in sert a pretty big capacitor in series with the 680 ohms but this is not a good idea. You can't lower it further without adding any trimming network. Maybe the best solution is to change the LM3875 but then you may mess up the board in the process.

Thanks.

I have measured it with the inputs shorted, the high side reads 57.6mV and the low side reads 13.1mV. If it's down to the chip I'll get a replacement, would it also be worth getting premium components with tighter tolerances?

I can't see any other way to remove the chip apart from cutting the pins, even removing an LED from a board by the "see saw" method is a pain.
 
Hi quickshift, offset on one of my amps is 120mv and I have no real issue with that. ;)

To quote good ol' Carlos:
To avoid high values of DC-offset, sometimes very different from channel to channel, you need to use the Ci caps (Look at Fig. 1 on the LM3875 datasheet).
Good electrolythics bypassed with quality poly 0.1uf caps will bring DC-offset to a couple of mv, without affecting the sound.

more here: http://www.diyaudio.com/forums/showthread.php?postid=469773#post469773
 
quickshift said:
I have measured it with the inputs shorted, the high side reads 57.6mV and the low side reads 13.1mV. If it's down to the chip I'll get a replacement, would it also be worth getting premium components with tighter tolerances?
In other words you have 1.8 mV and 0.41mV in input offset voltage (pretty good) but you have also a current flowing into the non-inverting input of the amp (see datasheet about input bias current). It's this current which adds on the extra offset.
 
peranders said:

In other words you have 1.8 mV and 0.41mV in input offset voltage (pretty good) but you have also a current flowing into the non-inverting input of the amp (see datasheet about input bias current). It's this current which adds on the extra offset.

Thank you.

OK, that's not so bad then.

I'll put the fire extinguisher away now. ;)

Actually, joking aside, such is my faith in my abilities with electronics that the first time I switched the amp on I used a bit of wood at arms length while holding a fire extinguisher across my face in case it caught fire or exploded. :( Mind you I made the loom for my track bike but that's just wires so it's easy.

Thank goodness people have had the time, skill and good sense to make kits so people like me without the ability to create an amp from scratch can also enjoy them.

Three cheers for the kit suppliers :)
 
Hello again.

Right, I have looked at the diagram, it's very nice but I don't understand it. :(

I have looked up bypass caps and it says connect the cap between the power and the main ground - that seems simple enough in itself.

I'm assuming I need to buy 0.1 uF poly caps and 22uF electrolytic caps - one per amp board but what do I do with them?

I have the rev 3 boards and haven't implemented the zobel.

Sorry for not really understanding much about all this stuff yet.
 
Sorted. :)

Right, I have worked out where to put the 22uF caps - I unsoldered ground side end of R3 (681 ohm) and lifted it from the board a bit, popped the - of the cap in the hole and then soldered the free end of R3 to the + side of the cap.

Results...

DC offset is now 1.7, 0.5, 0.5, and 0.5 :D YAY!!!!

I did 1 ch first and had a listen for a bit, adding the cap has further smoothed the sound, there was a tiny touch of grittiness or edge on the channel without the cap compared to the channel with the cap, I didn't notice it before but when switching between l + R on the preamp it was obvious. It's certainly made an improvement to the sound without any negative side effects. I'm assuming this is because the speakers have to work against the DC as it will slightly push the cones and offer more physical resistance in one direction of travel.

Many thanks for the info and help everyone. :)

I have also made the caps and snubber section of the carlosfm PSU (using a diagram!) from the decibel dungeon site (excellent BTW) and will add that between the PSU and amp boards next. :)
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.