LM3915 Issues

[MarcusS]

Still Trying are you saying to ditch the isolated dc dc converter all together.  Chris Mr, the -12V does already exactly sit at 18V and never budges with respect to the 36V without any capacitors.  The issue I had was crosstalk between the channels from an lm3915 point of view due to sharing the same dc dc converter.  I could take the signal from the input side of the amp but I do want it to be a calibrated db gain, THD and rms power meter.

[Chris Mr]

Yes, the -12 will 'sit' at 18v but only because of the two sets of input resistors to the LM3915 (R3, R4, R5, R6).  The 1nf capacitor on the DC DC converter (C13) provides a little smoothing but not much.  If you take a scope from the ground to the -12v to the LM3915 whilst there is the same audio present on both channels what do you get?  It looks like you would get the audio on the -12v (filtered a bit by the 1nf (C13) and the resistors (R3, R4, R5, R6)).

Are those polarised capacitors (C13 & C54)?

To put it another way - what forces the LM3915 -12 to be 18v?

[MarcusS]

No they aren't polarized capacitors they are just monolithic ceramics.  I had desoldered them earlier in the day to see if it would change anything.  Not sure why the -12 sits at 18V with respect to the 36 it just does and doesn't seem to budge. 

[Chris Mr]

.. and with a scope from GND to -12 and the same audio on both channels (full output) - what do you see?

A 'work around' would be to put a big (1,000uF) cap in place of C13; check polarity 

Trouble with that being that with low frequency audio going in one channel it would be 'seen' on the other if there were no input (depends on value of large capacitor).

[MarcusS]

Would a second isolated dc dc converter for the other channel lm3915 be the most technically correct solution and the wiring kept the same as the schematic the only difference being that each v- and each v+ for each lm3915 come from different dc dc converters.

[MarcusS]

I did try something similar to your work around already but with 1uF and I got all leds lighting up solid.

[Chris Mr]

Without a solid reference for the LM3915 -ve rail, no amount of isolated DC converters are going to help.

Assuming you don't have an oscilloscope, try connecting a 2K2 resistor between -12 and GND and then measure the voltage across it - over!

[MarcusS]

I checked on the scope and yes the 12- from the dc dc converter is moving around in response to the audio signal compared to the 36V gnd. However it does look like the leds are doing what I need them to do with the exception that it lights up leds on both channels.

[MarcusS]

Actually the leds don't really work properly.  I need 5V RMS for the first led to come on and then the last one comes on at 20V RMS.

[StillTrying]

The voltages on all 4 speaker connections move independently you can't common any of them, so I don't think there's any way you can do it that way without having 2 floating supplies. But...

I need 5V RMS for the first led to come on and then the last one comes on at 20V RMS.

Think of it this way, I think it's accurate enough if you want to go in this direction, if you want to measure at the speaker connections then I don't think you have a choice.

At 0dB no-output the amp holds both ends of the speaker at the mid voltage point, - both ends at 18V on 36V supply, so that there's 0V across the speaker.

Your 20V RMS is peaks of +/-28V, so..

To get this peak of 28V across the speaker connections the amp raises one speaker connection 14V above the mid voltage point, and the other speaker connection 14V below the mid voltage point, giving 28V across the speaker. As the voltage on each side of the speaker is an exact +/- mirror image of the other side, you only have to look at one of them referenced to 0V.

Looking at just this one side of the speaker, at 0dB output the speaker is held at the mid point = 18V , when the 28V peak comes along this one speaker side raises to 18V + 14V = 32V.

This gives you the 2 Vref levels needed for the LM3915, Vref LO = 18V and Vref High = 32V, referenced to 0V.
In practice you'd potential divider this voltage down, div by 10 gives Vref Low = 1.8V and Vref High = 3.2V referenced to 0V.

[MarcusS]

Does pin 4 R Low need to connect to pin 2 on the lm3915? If it doesnt then couldnt we put spk + and 36v ground between 5 and 4 with the appropriate divider and power the 3915 between pin3 and pin 2 with the dc dc converter.

[StillTrying]

Does pin 4 R Low need to connect to pin 2 on the lm3915?

Yes, if you want to ref everything to GND, the TDA7498's gnd pin4, the LM3915's gnd pin2, and the DC-DC converters -ve can be connected together.

The LM3915's Ref Low doesn't connect to this gnd because you're now detecting one side of the speaker's voltages, 18V = 0dB, 30V = 30dB. If you supply this 18-30V to the LM3915's Sig In via a div 10 potential divider the Vrefs you need are then Vref Lo = 1.8V and Vref High = 3.0V, they're handy values of Vrefs because the difference between them is 1.2V, which is the native voltage of the LM3915's voltage reference = no voltage adjusting resistors needed.
You'll still need 1 resistor from Vref High(3.0V) to gnd to set the LEDs current.

Is that the direction you're trying? I'll update the sim with these values soon.

[MarcusS]

I cant connect the Vout - of the DC DC converter to 36V gnd.  But I take it from your response that I need to connect pins 2 and 4 together on the LM3915 for the IC to work.

[StillTrying]

I cant connect the Vout - of the DC DC converter to 36V gnd.

Oh dear! Why not when it's isolated.

But I take it from your response that I need to connect pins 2 and 4 together on the LM3915 for the IC to work.

No pin 2 is it's power gnd and pin 4 is Vref Low.

I've worked out some values(or rather LT has) as below, I've had a play with it and will add description as needed, can't you try something like that?

R5, R6 and C4 produce Vref Low from the speaker's mid-voltage, Green on the plot.
The LM3915's internal 1.28V ref produces Vref High = (VrefLow + 1.28V). Red on the plot.
The Sig In in Blue on the plot shows a 18W peak 500Hz signal with its +ve peaks crossing the vref Low value.


Edit:
I wound up the 500Hz signal until it's peaks just reached the Vref High LED #10 level, Watts in the 4R speaker were 250 peak, 175 RMS, that's not far off.

[MarcusS]

When i connect vout - of the dc dc converter to 36v gnd all leds go hard on.  In your diagram is 0v the vout - of the dc dc converter and 12v = vout+ from the dc dc converter.

[MarcusS]

Also, if pin 2 and 4 dont need to be connected then i could connect pin4 to 36v gnd pin 5 to the output of the amplifier and then the dc dc converter connects between pin 3 and pin 2.  Would that work also.

[StillTrying]

When i connect vout - of the dc dc converter to 36v gnd all leds go hard on.

You must have something else connected somewhere such as the Vrefs.
Careful you don't blow the LM3915, 3V to 25V max.

In your diagram is 0v the vout - of the dc dc converter and 12v = vout+ from the dc dc converter.

Yes, the 0V and 12V power across the Lm3915 is supplied by the DC-DC converter.

[MarcusS]

The aim of the vu meter is for the last led to indicate clipping (22v rms = 80w rms) and then each led lower is 3db less. I.e. 40,20,10,5,2.5,1.25,0.6,0.3,0.1.

[StillTrying]

The aim of the vu meter is for the last led to indicate clipping (22v rms = 80w rms) and then each led lower is 3db less. I.e. 40,20,10,5,2.5,1.25,0.6,0.3,0.1.

22V RMS across a 4R speaker is 121W RMS. Have I got the wrong speaker R value?
OK I'll use 6R.

22V RMS is 30.8V peak, but Vref High in the sim is 34V(after the X11 potential divider) so I'll try to tweak it down a bit.
LTspice is quite easy to use when someone is providing the schematic, if you decide to have a try of it.

[MarcusS]

6 ohm speaker system i have designed to pair with this amp.

[MarcusS]

I will give it a go.  Im trying to get my head around why your circuit has a voltage divider for div low instead of that being connected to 36v gnd. The other voltage divider makes sense although my calcs give me 3.3k and 27k.

[Audioguru]

TDK company does not make a CCG30-4812s voltage converter. A voltage converter will have the +36V and ground input and have a +12V and ground output. The grounds are connected together and there is no -12V that is not needed.

Of course the outputs of the BTL amplifier are at about half the positive supply voltage but the input of the LM3915 must be averaged at ground so simply use a coupling capacitor from one output of the amplifier and feed it to the input attenuator of the LM3915.

Your LM3915 has no pin numbers.
Why does your circuit have a "Vref Lo"?? Pin 4 of the Lm3915 should be connected to the circuit ground.
Why is your Vref so low? Make it about +10V (and adjust the input attenuator to match) then input offset voltages will not mess up the outputs.   

[MarcusS]

Tdk does make a ccg30-48-12s because they are soldered and working on my board in front of me.

[Audioguru]

Actually the leds don't really work properly.  I need 5V RMS for the first led to come on and then the last one comes on at 20V RMS.

You do not understand that the LM3915 has a logarithmic response. The datasheet shows that if the 10th LED lights with an input of +10V then the 1st LED lights with an input of +0.447V. Each output is active with 0.707 times (-3dB) less level than the next output. Then with an input of 20V RMS to light the 10th LED, the 1st LED lights with an input of 0.894V RMS.   

[MarcusS]

You must also understand that most dc dc converters are isolated and thus the output gnd is not connected to input ground. If it was it would be non isolated and this thread wouldnt have started. You can search on farnell or digikey to see that there arent many non isolated converters and for high end audio you dont want other voltage sources being able to put noise onto the main amplifier gnd anyway.