LM3915 Issues

[MarcusS]

Has anyone else noticed the datasheet for the LM3915 Audio Power Meter Schematic on page 26 is wrong.   
I'm trying to add a nice looking Vu meter to the output of an amplifier I have been designing and got myself caught up on connecting the speaker minus output of the amplifier to the ground of the vu meter circuit basically shorting out one side of the audio amplifier.  The calculations for the voltage divider are simple enough as is calculating Vref but that all goes out the window if you cant connect speaker minus to pin 2 and pin 4 of the IC.

I have had it working but incorrectly by connecting the audio input between pin 4 and 5 but the problem is pin 4 is no longer referenced to pin 2 of the IC.  Has anyone built 2 of these for stereo run it from a single voltage supply and not shorted out the audio amplifier or created crosstalk between channels.  I am keen to see if anyone has done anything fancy with resistors or decoupling capacitors to get around the issue.

[rollatorwieltje]

The TI datasheet? It's not wrong, it just assumes the power supply to your LM3915 is floating. You should power it from a separate power supply, i.e. another secondary winding on your transformer or an isolating DC-DC converter.

[MarcusS]

My lm3915 circuitry is actually being powered off a recom rp30-4812 which is a shielded isolated dc dc converter but the problem is the 12v minus out of the dc dc converter is shorting out speaker minus of the btl amplifier design as 12v minus out is still essentially ground in comparison to speaker minus which is the negative side of the amp output waveform.

[Audioguru]

The ground of a BTL amplifier is the ground for its audio. The speaker does not connect to ground since both of its wires are fed signals with opposite phase.
The LM3915 power meter circuit does not connect to the -12V, it connects to one of the wires feeding the speaker and the amplifier's ground. Ground is the 0V between the +12V and the -12V.

[MarcusS]

I cant understand how you could connect it that way.  The lm3915 is being powered off an isolated 12v dc dc converter it has a vout+ = 12v and a vout- = dc dc converter output ground.  The dc dc converter ground has a significant voltage differential to the amplifier ground.  I have tried connecting one side of the amplifier output to the input and connecting pins 4 and 2 to the vout- of the dc dc converter but it doesn't work correctly. 

I am happy to follow instructions if someone can tell me exactly how they would connect 2 lm3915s (1 for each channel) powered off a single TDK CCG30-4812s with the outputs of each channel of a TDA7498 BTL class d amp IC going to each lm3915.

[Audioguru]

Connect the input of the LM3915 circuit to only one BTL speaker wire and the amplifier's ground. Then re-calculate the value of R1 for half the output voltage swing. Then the LM3915 circuit can use the amplifier's positive and ground power supply.

[MarcusS]

Where do I connect the Vout- of the dc dc converter too then.  It doesn't make sense. The datasheet says to connect pins 2 and 4 together which is connected to ground and feed the input into pin 5.  I cant connect the vout- of the dc dc converter to amplifier supply ground.

[void_error]

Why do you need an isolated DC-DC converter in the first place?

[MarcusS]

There are 2 reasons.  I have the converter on the board for other things and secondly the LM3915 will not handle the 36V that powers the amp.

[Audioguru]

Reduce the amplifier supply voltage or add a new positive one so that the LM3915 is powered from about +9V then it will not get too hot when powering a bar of 10 LEDs.

[MarcusS]

I have a positive supply voltage for the lm3915 it is called a 12v output dc dc converter.  It takes the 36V input that powers the amplifier TDA7498E and reduces it down to 12V which I need for some other things and happens to be the ideal voltage for a 10V logarithmic display using an LM3915.

I cant lower the amplifier supply voltage, it is a 160Wx2 BTL amplifier and it nominally requires 36V.  I don't have a problem powering 10 leds and the lm3915 from the dc dc converter I have a problem with ground references.

I take it that I can only try to find a non isolating dc dc converter module so that the 12v ground from the dc dc converter is the same as amplifier ground.  At the moment, the dc dc converter output ground is 18V above the amplifier ground.

[MarcusS]

Can't exactly find any non isolated dc dc converters that meet me requirement.  Don't really want to have to put on the board a non isolated and an isolated dc dc converter.  Has anyone actually powered an lm3915 from an isolated dc dc converter and knows what needs to be done.

[Audioguru]

Here is as tutorial for a DC to DC buck voltage converter. It uses PWM to reduce the voltage and the input ground, circuit ground and output ground are all the same and can be the same as the amplifier and LM3915 grounds. https://www.maximintegrated.com/en/app-notes/index.mvp/id/2031

If the supply for the LEDs is +12V and the LEDs use 2V then the output transistors in the LM3915 have 10V across them. If the LED current is 20mA and all 10 LEDs are lit then the poor LM3915 must dissipate 200mA x 10V= 2W but its maximum allowed dissipation is only 1.3W when it is very hot. So you need fewer LEDs lighted or less current or less voltage.

[MarcusS]

It doesn't matter about the maxim application note.  I have not built a buck converter for this application as I need EMC compliance.
My DC DC converter modules do not have input and output ground connected together.  As I said before I have a TDK CCG30-4812s.

I have calculated and measured my LED current as 7mA each my vref is made of 3.9k and 22k which if you use the formula in the lm3915 datasheet gives an led current of 7.5mA. This also aligns with what I am measuring.  That combined with one 82 ohm resistor in series with the common of all leds gives only 200mW dissipated in the LM3915.  So 12V is not and never has been an issue.

I don't have an issue calculating the values for an LM3915.  My issue the whole time has been how to handle completely isolated supply voltages in the system that DO NOT share a common ground.  If I had a common ground I wouldn't even be asking the question.

The problem I have just discovered is that the 2 LM3915s must not share the same ground i.e. the same output gnd of the dc dc converter.  I have just proven that what I was seeing is that the leds are responding to left minus right and that is showing on both channels of lm3915 even if only one channel is driven.

[void_error]

Hmm... since it's a BTL amplifier with a single supply then the L+/L- & R+/R- speaker outputs will idle at roughly half of the supply voltage, and with an input signal present they outputs will be 180 deg out of phase - L- is L+ inverted.
You only care about the the L+ and R+ for the amplitude.

Looking at the LM3915 datasheet I can see the input is internally clamped to the negative supply which means it'll act almost like a half wave rectifier.
If you AC couple the input of the LM3915 you can tie the negative output of your DC-DC converter to the ground of the BTL amp but since you said you can't do that this is not a solution.

What you can do is design (or just buy it, but they're expensive) an isolation amplifier using optocouplers, a google search will give you an idea how, and reference their inputs to the amp's ground. That way you isolate the input of the LM3915.

[StillTrying]

I think this is quite a tricky problem.
Is the 36 V supply quite stable?  It's not a class D is it? 

This is what an hours thinking came up with, I've not done the values, or checked the impedance needed for driving Ref Lo, etc. etc... Just 1 channel shown.

[void_error]

It's not a class D is it? 

It's a TDA7498, which is a class-D amp.
His issue is that he can't reference the LM3915 ground to the amp's ground, that way it would have been a lot easier.

[StillTrying]

It's a TDA7498, which is a class-D amp.
His issue is that he can't reference the LM3915 ground to the amp's ground, that way it would have been a lot easier.

I think he can as long as he can take the 0dB being the mid ~18V point.

Here's an attempt at a class D version, but only peak detects on 1/2 of the waveform.

[MarcusS]

I was just mucking around with some diodes myself before I saw the posts.  At the moment I am contemplating adding a second isolated 12v dc dc converter for the other channel or option b although still haven't gotten it confirmed in my head if I put diodes between V- of both lm3915 ic's so that they cant backfeed through each other.

[MarcusS]

If only 1 channel is driven the leds are correct in terms of magnitude but the issue is the leds light up on the other channel.  If I drive both channels it seems like the leds only show up when there is a difference between the channels. 

If I use a signal generator on both channels with the same frequency and magnitude no leds light up at all on either channel.

[StillTrying]

although still haven't gotten it confirmed in my head if I put diodes between V- of both lm3915 ic's so that they cant backfeed through each other.

As each channel has an independent signal I don't think there's any way to join their 0V/grounds, 2  independent floating supplies are needed. Even if you go that way, you're still only going to be looking at peaks of one polarity aren't you, Is looking at the peaks of just one polarity good enough?

If I drive both channels it seems like the leds only show up when there is a difference between the channels. 
If I use a signal generator on both channels with the same frequency and magnitude no leds light up at all on either channel.

We could do with a schematic.

[MarcusS]

Yeah I only need to see one side of the waveform.  Ill put up the schematics.

[MarcusS]

Amplifier Schematic.

[StillTrying]

I still think this is still the way to go, no isolated supply needed. I made it into a sim working version 250kHz ?, and fiddled with R1,2,3,4 to make  Ref_Hi - Ref_Lo = 1.2V.

LT worky version included if NE1 wants to improve!

[Chris Mr]

The -12v from the LM3195s' DC DC converter is floating with respect to the audio - there is no other audio reference to the LM3915s (as far as I can see).

A suggestion would be to connect the LM3915 -12v via a pair of resistors to 1/2 of the 36v supply.  It would need a large cap so the two audio inputs don't pull it around DC wise.

In other words, one (say) 3k3 resistor from +36v to -12 (from DC DC converter) and the other from -12 to ground.  Then a large cap from the -12 to ground - say 1,000uF (needs experimentation at low volumes).

Another idea would be to take the LM3915 input from the audio input (wiper of volume control) and then connect -12 to ground.  That relies on the audio input being AC coupled as it comes in (no DC offset on the input).