[Jamdat]

tl;dr Modified a CMoy with TLE2426 which gets super hot. Normal or horrible beginner mistake? Question at bottom and tests/explanation in between.

Hello Head-Fi forums. I've been lurking around here for a while, but now I need to actually ask for help.

I had a need to listen to TV with headphones and this led me to the CMoy amp. I settled on Tangent's helpful tutorial and diagrams since I am coming into this with little electrical knowledge. After buying all the components, I proceeded to royally screw things up by reading the diagrams backwards, getting frustrated with my iron, putting things in backwards and upside down, and completely frying things. It turned out to be such a mess so I threw it all out and started again.

I've done extensive reading of pretty much every page on Tangent's website. I spent lots of time looking at the discussion on virtual grounds(Here). For my second attempt, I decided to slightly modify the CMoy design to incorporate a better ground. So, my design now has a 24V DC rated power source (unregulated) running through a 7824 regulator (the DC adapter puts out 30.2V with no load, so I am above the dropout voltage of the regulator). I then have a TLE2426 (TO-226 3 pin package) splitting the voltage equally. Then I have the op-amp hooked in. The schematic that I have created is below. Yes, that's right, it's just the TLE2426 circuit from Tangent's page mashed into the CMoy circuit like the total beginner I am. I added the 7824 circuit from the datasheet for the part. Is it the best, I have no idea. Does it need more/less components? No clue.

​

The schematic diagram for my modified CMoy.​

I've laid everything out on a breadboard to test it. I used my computer for a sound source. I plugged the adapter in, clipped alligator clips onto my headphones and sweet, sweet music came pouring out. My mashing in of the TLE2426 worked. I was very happy, listened to an entire song and then... burned myself on the TLE2426!! Holy crap is that thing running super hot!

​

An overview of my breadboard. It is on right now as indicated by the LED.​

 ​

I got worried, so I did some playing around. I first checked every connection and verified that it was laid out correctly. What I found was that, with no headphones hooked in, nothing got hot at all. Input DC voltage was 30.1V. The voltage coming out of the 7824 (measured from the 'out' pin to the '-' pin of the DC jack) was 23.7V (exactly what it should be). The 7824 was not hot (maybe a bit warm, but barely). The TLE2426 split the voltage exactly down the middle (like it should). Between the 'IN' pin and '-' on the jack was 23.7V. Between the 'OUT' pin and '-' was 11.85V. Perfect. So with no sound hooked up, everything is working out alright.

I then did some experiments with sound on and a multimeter. What I found was that as soon as the sound started, the voltage between the 'OUT' pin and the '-' pin of the DC jack (which was previously 11.85V) was now 19.1V. This got me thinking that there was too much voltage and the TLE2426 was getting hot because of it. So I did some more experiments where I put the ground connections in different places.

Some terminology: I have a 3.5mm cable plugged into my computer. There are three clips attached to it, L, R, and Ground. L and R go into the inputs of the amp. I have called the cable connected to this ground 'Gcp'. The L and R output of the amp is clipped to my headphones. There is another cable clipped to the ground part of my headphone plug. I have called this 'Ghp'.
Gcp = ground from computer jack
Ghp = ground clipped to headphone plug

​

How I connected to my computer. Black=Ground(sleeve). Yellow=Right(ring). White=Left(tip). Gcp is the black connector here.​

 ​

​

How I connected to my headphones. Black=Ground(sleeve). Yellow=Right(ring). White=Left(tip). Ghp is the black connector here.​

Now the results of my test:
Gcp disconnected, Ghp connected to Vground: Loud buzz.
Gcp disconnected, Ghp disconnected: very very faint sound. Op-amp is working because it starts crackling when I raise the volume. Op-amp gets hot. TLE voltage is 11.85V.
Gcp disconnected, Ghp connected to '-' DC jack: No sound. op-amp got super hot really fast.
Gcp connected to Vground, Ghp connected to Vground: TLE2426 gets very hot. TLE voltage is 19.3V.
Gcp connected to Vground, Ghp disconnected: Faint sound. Op-amp warm. TLE voltage is 11.85V.
Gcp connected to Vground, Ghp connected to '-' DC jack: No sound. Op-amp got very hot.
Gcp connected to '-' DC jack, Ghp connected to Vground: TLE very hot. Voltage 2.9V.
Gcp connected to '-' DC jack, Ghp disconnected: TLE and op-amp ever so slightly warm. Voltage 11.85V.
Gcp connected to '-' DC jack, Ghp connected to '-' DC jack: Op-amp gets hot. Also sound only comes out one channel.

Judging by my testing, it seems the best is that there should be no ground connection going to the output. But, even with my limited electrical knowledge, this makes no sense to me.

I then hooked up a 9V power adapter (7824 falls out of regulation and has an output of about 7.7V) With this voltage source, nothing gets hot, but the voltage between the TLE 'OUT' and the '-' DC jack is 4.44V instead of the 3.85V like it is with no headphones connected. Everything else is fine (it even sounds ok). It seems that with both power sources, the TLE voltage for common ground (which should be half of the input) jumps up by a big chunk as soon as I plug things into Vground.

​

Blue stripe on bottom=V-, Red stripe on bottom=Jack positive (30V), Blue stripe on top=24V from regulator, Red stripe on top=Vground. Brown jumper carries the regulated voltage.​

 ​

​

Different view. Right hand white and yellow clips are inputs on pins 3 and 5 of op-amp. Right hand ones are outputs on pins 1 and 7. The optional resistor has been omitted.​

 ​

​

Final view. Just for completeness in case somebody can spot an error in wiring.​

Question:
Rounding out my rambling post into a question for more experienced members: Is the TLE2426 getting extremely hot a normal occurance, am I hooking things up incorrectly, or am I doing things really wrong in the schematic? Another option is that my 30V regulated to 24V is just too much for this amp.

Sorry about the long post, but I wanted to give lots of information.

Thanks in advance for your help.

 

[Zigis]

Your schematic's amp part is wrong. If you want amp with gain, you need resistors between chip's out and -in, you have bridge there, see original shem. If you want unity gain buffer only, bridge output to -in, and nothing to ground. You have something between, this is wrong. I don;t now are this TLE heating reason
 
Normaly TLE is cold, 24V is good for this amp.
 
BTW, your alligators looks very dangerous animals. At least solder 3 wires to mini jack for input and 3 wires to 3.5mm socket for output, for your future experiments.

 

[Spacehead]

TLE2426 gets hot if it gets shorted.
 
You have one or both the rails shorted to TLE2426 virtual ground if it gets hot.
 

 

[bmwpowere36m3]

Agree, either is shorted internally and toast or you have it mis-wired.

 

[tangent]

Are you powering the amp from the laptop?  If so, that's likely your problem.  The laptop power source's V- pin is probably tied to audio ground inside the laptop.  If so, you don't have an isolated power source.
 
This issue is covered in the CMoy tutorial here.  For more, search this forum for USB, car, and PC power threads.  It's all been covered before.

 

[Jamdat]

This post actually disappeared and so I didn't expect any replies. I've done some more experiments since I wrote this a week ago and I think I've narrowed down the problem.
http://www.head-fi.org/forum/thread/492616/do-laptops-cause-problems-with-cmoy
 
Quote:

Are you powering the amp from the laptop?  If so, that's likely your problem.  The laptop power source's V- pin is probably tied to audio ground inside the laptop.  If so, you don't have an isolated power source.
 
This issue is covered in the CMoy tutorial here.  For more, search this forum for USB, car, and PC power threads.  It's all been covered before.

No, I'm not powering it from the laptop. I'm using a 24V source (wall wart). It's unregulated, and your tutorial mentions that all unregulated sources are isolated.
 
In the week since, I've found that this issue only arises when I plug into my laptop. I rigged up a cable which plugged into the back of my TV and when using that as a source, there is no problem, no voltage imbalance, and no heating. Using a desktop computer as a source leads to a buzz in the headphones, but no heating or voltage imbalance. Oddly, the buzz goes away when I touch the 220uF capacitor. Same thing using the headphone jack on a portable stereo - buzz, but no heating and no voltage imbalance.
 
Given this other info, I think that I've wired everything correctly, and neither the TLE, nor amp, is broken. I think that the laptop is causing this issue, and I'll be researching more about that on here.
 
Any thoughts?
 
Thanks

 

[Fred_fred2004]

Have you tried running it off a battery?

 

[tangent]

Quote:

I'm using a 24V source (wall wart). It's unregulated, and your tutorial mentions that all unregulated sources are isolated.

 
Let's test it anyway.  Take your DMM, put it on continuity mode (or the lowest resistance range, if it doesn't have a continuity test mode) and see if you get continuity between any input and output terminal.  With a 2-blade input and just two output lugs, that means you should end up with four test results.  If the AC side has 3 lugs, there are 6 combinations.  None of these tests should give you anything but a very high resistance, if you get any continuity at all.  If you get a low resistance, it's not isolated.
 
I'm concerned by the "+/- 24 V" shown on your schematic.  If true, that's awfully wasteful given the 7824 regulator -- you're throwing away at least twice the available power in heat just in this one part.  If that's not actually the case, then it speaks of a confusion that could have lead to any number of incorrect implementation choices.
 

 

[Jamdat]

Quote:

Have you tried running it off a battery?

I have, with the same results. Although the TLE doesn't get hot, the voltage still becomes unbalanced. From a balanced voltage of +3.8/-3.8, it becomes +0.6/-7.1. Sound only comes out one channel.
 
This problem doesn't occur when I use my TV as a source - with battery or wall power.
 
Quote:

Let's test it anyway.
...
With a 2-blade input and just two output lugs, that means you should end up with four test results....

Good advice, assume nothing. I've just done this: I have 2 prongs at the input and I tested all four combinations. No continuity between anything. That should mean that it's isolated.
 
Quote:

Originally Posted by tangent /img/forum/go_quote.gif
I'm concerned by the "+/- 24 V" shown on your schematic.  If true, that's awfully wasteful given the 7824 regulator -- you're throwing away at least twice the available power in heat just in this one part.  If that's not actually the case, then it speaks of a confusion that could have lead to any number of incorrect implementation choices.

Confusion is possible here. I used your schematic as a reference and replaced the two batteries with the power adapter since that is what you would do if you followed the 'wall power only' advice here. My power adapter is centre positive, so I put a '+' symbol on top and a '-' symbol on the bottom. The regulator keeps everything at 23.7V since the '24V' adapter actually puts out more like 30V.
 
I made this schematic before I realized that a power supply puts out one voltage and the '-' terminal is just for return. I had originally thought that, since you used two batteries, one was providing '+' and the other was providing '-'. It only dawned on me that it was otherwise when I pondered how people ran these things with only one battery. So, if you replace the +/-24V symbols with a giant 24V battery, that's what I was trying to get across in my schematic. I didn't know the correct symbols for a power adapter, so I just used a 'supply' symbol. There isn't a 48V difference between the terminals, and this is probably very poor labeling on my part. Is this what you were concerned about?
 
The implementation that I drew up makes lots of sense to me (now) and seems like it should work perfectly. Indeed, it does; with the exception of this small laptop issue. Are you saying that it's actually incorrect and it's been working just by pure chance? I don't think that is so...
 
I've done a lot of learning by troubleshooting this project. It's almost more enjoyable that building it.

 

[hiker101]

It's very suspicious that the laptop behaves differently than your other sources.
 
Is there a DC offset in the signal coming out of the laptop?
 
Check to see if there's a DC voltage between any of the three input signals coming from the laptop. (R/L/G)
 
If you're hearing a "buzzing" sound maybe the laptop has pushed your amp into oscillation (perhaps with excessive noise in the MHz region).  This might show up as a DC offset at the output- bad for headphones if it's big enough.
 
Have you checked your amp for a DC offset at the output with no source?  With inputs shorted, measure the V diff between R/L/G out.

 

[Jamdat]

Quote:

It's very suspicious that the laptop behaves differently than your other sources.

I know! I've just tested it with an iPod and there is, again, no problem.

Is there a DC offset in the signal coming out of the laptop?
 
Check to see if there's a DC voltage between any of the three input signals coming from the laptop. (R/L/G)

This part has always confused me, so I'm going to ask if my method of measurement is correct.
I've plugged a plug into the headphone jack on my laptop and I have three alligator clips attached to R,L, and G. I place my meter in 2000m DC mode and clip the ground (COM) port to the G of the headphones. With the other tester, I touch the R or L pin.
Doing this, I get about 10mV between G and R, and about 12mV between G and L. I get between 0 and 3mV between R and L. I think it fluctuates because it's such a small reading.
Is this the correct way of testing?
 

Have you checked your amp for a DC offset at the output with no source?  With inputs shorted, measure the V diff between R/L/G out.

With no source, I shorted the inputs (and by shorted I mean put a wire from the L input to the R input) and measured the voltage between L,R, and G (this is the virtual ground). 0V between L and G, 0V between R and G, and 0V between R and L.
 
Still puzzled here.
 

 

[Zigis]

Quote:

Quote:

With no source, I shorted the inputs (and by shorted I mean put a wire from the L input to the R input) and measured the voltage between L,R, and G (this is the virtual ground). 0V between L and G, 0V between R and G, and 0V between R and L.
 
Still puzzled here.
 

Shorted inputs actually mean each channel inputs L and R connected to ground, not L to R.
Something is wrong with your measurements, there is always DC offset on output, few mV or even few tents of mV with good chip.
Problems is wen DC is higher then 20-30 mV.
 

 

[Spacehead]

Wouldn't you already do it on perfboard? Schematic looks right. There must be a short somewhere. 

 

[Jamdat]

Shorted inputs actually mean each channel inputs L and R connected to ground, not L to R. Something is wrong with your measurements, there is always DC offset on output, few mV or even few tents of mV with good chip.
Problems is wen DC is higher then 20-30 mV.

Hmm, so I did measure incorrectly.
With both inputs connected to ground (Vground), I still get close to zero (0.1 and 0.4 mV). I'm going to put in a new chip because if there's always DC offset, then something is clearly wrong here since I have the 2132P, which seems to be an ordinary and not a 'good' chip. I'll update with new numbers with the new chip.
To clarify, with power connected but no source, I connected the inputs to ground, and then measured from ground to the output.
 
Sigh...
 
Spacehead, I decided to use a breadboard before soldering things because I knew I would make these sorts of mistakes. A lot easier unplugging wires than it is resoldering. I thought the schematic looks correct as well and I am tempted to just solder the thing - it works with the TV and that's what I built it for - but it seems like the easy way out.

 

[tangent]

Quote:

No continuity between anything. That should mean that it's isolated.

 
Yes.  That and the 30 V under no load condition measurement reassures me that it's truly unregulated and isolated.
 
Quote:

I didn't know the correct symbols for a power adapter, so I just used a 'supply' symbol.

 
There are several, varying depending on the type of plug you're using.
 
Unfortunately, the EAGLE libraries aren't really set up for DIY projects, so to get a barrel jack symbol for the schematic, you'd probably have to pick a barrel jack for the PCB side.  Not a big issue if you're only trying to draw a schematic, but if you also want a PCB, you end up having to build a custom part, grabbing the barrel jack symbol and attaching it to the wire pads or whatever you want on the PCB side.
 
This is one of the few virtues of ExpressPCB relative to EAGLE: there isn't a hard tie between schematic symbol and PCB component.  You can mix and match at will.  You only have to draw a given schematic symbol once, then you can use it with any PCB footprint that provides the right set of corresponding pins.
 
Keep the original purpose of schematics in mind: to communicate information about a design to other engineers.  The designers of some CAD tools don't quite get that, thinking the primary purpose of the schematic is to communicate the design to the tool.  It can be anywhere between difficult and impossible to produce a really good looking schematic in such a tool.  I'm not singling out EAGLE here.  Every SPICE schematic editor I've used is even worse.
 
I redraw most of my EAGLE schematics in gschem for that very reason.  I develop the PCB with the EAGLE schematic editor, then when the schematic is finalized, redraw it in gschem for readability.
 

Quote:

There isn't a 48V difference between the terminals, and this is probably very poor labeling on my part. Is this what you were concerned about?

 
Yes.
 
Quote:

I've plugged a plug into the headphone jack on my laptop and I have three alligator clips attached to R,L, and G. I place my meter in 2000m DC mode and clip the ground (COM) port to the G of the headphones. With the other tester, I touch the R or L pin.

 
The alligator clips are just making things difficult here.  I just hold the COM lead from the DMM against the headphone plug's sleeve, then touch the mV lead to tip and ring.
 
I recommend alligator jumpers for the audio connections in the CMoy tutorial because there are six of them, too many to hold in place with two hands, especially since those two hands will also be needed to fiddle with switches and knobs during the tests that follow.
 
Quote:

something is clearly wrong here since I have the 2132P, which seems to be an ordinary and not a 'good' chip.

 
Au contraire.  The OPA2132P is one of the better chips.  It won't win many audiophile listening competitions, but its DC performance is excellent, particularly the non-A variant.
 
Sounds like time for board pics.  Macro photos, both sides.  Use a tripod and lights, if you can.