Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Amp question. (Read 45281 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Amp question.

Reply #50 –
Quote from: HbG on
Bleh, and i promised myself i wouldn't spend more on audio for a while... I am proficient with a soldering iron but i'd still rather not take the risk. I just know what a hellish job it would be to replace all that.

Anyway, i would assume the only common mode of failure is changes in capacity (a reduction i suppose). What is the risk of the other horrors you mention? I never hear about it.

I opened up the amp and did a visual inspection. There is no bulging or leaking that i could spot. As expected there is one really huge cap for the powersupply and two big ones for the capacitively coupled output, they look suspiciously new compared to the rest of the parts. You can also see the four transistors of the main stage and the crowded preamplifier circuit board bolted in front of it.The amp also definitely belonged to a smoker, given the smell it has inside and the kind of greasy-ish dirt that is on the components (is there such a thing as a parts washer for electronics?). But the volume knob is absolutely fine so i'm not complaining.

Capacitors (electrolytic) tend to fail as open or short circuit or just by a reduction in value. Short circuit isn't nice - it can do untold electrical damage to other components. I've also seen equipment where the only evidence remaining of a capcitor was two leads sticking out of the board and the rest of the component splattered around the insides of the casing! Any bulging capacitor should be replaced without question. At that age I'd say replace them all, unless you aren't worried about the loss. I guess for the equipment price it may not be worth it financially, however.

If the amp uses output capacitors, as you say, I'd steer clear of bass-reflex designs of speakers too. The suspensions are liable to be damaged by the switch-on thumps as the capacitors charge (some amps have bleed-resistors and a speaker relay, delayed, to combat this).

Amp question.

Reply #51 –
Quote from: Gigapod on
Well, I can see you really caught the DIY virus! You can test the circuits using a protoboard, I think that would avoid a lot of problems with solder resin.

Anyways, since you passed me the vintage virus, here is what I ended up with, for the princely sum of 30 euros:
http://gigapod.free.fr/marantz_restoration.html


16Ohm headphones appears to be a difficult impedance to work with. I got it under control now but can't get rid of the clipping. It's much better with 32 ohm 'phones.
Chu Moy himself recommended me to drop the gain to 2, but i'll need to solder a new amp to do that. Indeed i should get a protoboard. I'll get some more parts tomorrow. That'll be number three...
Here are two pictures

That's a real nice Marantz you got there, too, even if not as "vintage". Congratulations.  Love the power meter. Thorough story on the webpage too. I wish i knew that much!

Antz: There is no switch-on thump. I'll have the caps replaced - at some point. Since it looks fine inside and sounds fine too i don't consider it a time-critical operation.
Veni Vidi Vorbis.

Amp question.

Reply #52 –
Quote from: HbG on
...
16Ohm headphones appears to be a difficult impedance to work with.
...

For an IC opamp, it is: their output stage is not really designed to drive these loads, and you get all sorts of problems: distortion, crosstalk, clipping (and if you had used an old TI op-amp you would see some horrible things when clipping, like polarity reversal and latching), thermal effects due to the small size of the die, a serious reduction in gain, etc.
Of course you can solve all that by adding a discrete class AB buffer (2 transistors, 5 or 6 resistors and two 1n4148 diodes; total cost < 1 euro), as suggested by Chu Moy himself(http://headwize.com/projects/showfile.php?file=opamp_prj.htm, figure 11).
However... the 9V battery will run down faster.
or you can use the TDA1308 chip http://docs-asia.electrocomponents.com/web...66b8002de34.pdf
which was designed by Philips engineers to solve all these problems.
Quote
...
Here are two pictures

Looks quite good to me.
Quote
That's a real nice Marantz you got there, too, even if not as "vintage". Congratulations.  Love the power meter. Thorough story on the webpage too. I wish i knew that much!
...

Thanks! I am quite happy with it. For the price I paid, it was a good buy.

Amp question.

Reply #53 –
Hmm, an output stage might be a good idea for a next design. The TDA1308 is nice but doesn't feel DIY enough.

Anyway the TS922 can drive 80mA of current, which in 16ohm headphones gives 102.4mW of power, more than enough. The culprit is actually the TLE2426 rail splitter IC i'm using, as it can only deliver 20mA and transients easily exceed that. I didn't put output caps on it as the circuit on audologica didn't show any, but apparently i should have. I wonder if these things can be soldered in parallel.
Veni Vidi Vorbis.

Amp question.

Reply #54 –
Quote from: HbG on
Anyway the TS922 can drive 80mA of current, which in 16ohm headphones gives 102.4mW of power, more than enough. The culprit is actually the TLE2426 rail splitter IC i'm using, as it can only deliver 20mA and transients easily exceed that. I didn't put output caps on it as the circuit on audologica didn't show any, but apparently i should have. I wonder if these things can be soldered in parallel.
You could solve the problem with transients by putting large caps on the output of that rail splitter. It'll spoil the compactness of the amp somewhat, but will make your rails much more stable.

Amp question.

Reply #55 –
Quote from: HbG on
Hmm, an output stage might be a good idea for a next design. The TDA1308 is nice but doesn't feel DIY enough.


  You have really been bitten by the DIY bug!!!

Quote
Anyway the TS922 can drive 80mA of current, which in 16ohm headphones gives 102.4mW of power, more than enough.


That's the short-circuit current, when driving very low impedance loads (servo applications and others). You'll get lots of distortion at these levels of current. But I don't think you'll need 100mW anyway.

Quote
The culprit is actually the TLE2426 rail splitter IC i'm using, as it can only deliver 20mA and transients easily exceed that. I didn't put output caps on it as the circuit on audologica didn't show any, but apparently i should have. I wonder if these things can be soldered in parallel.


A 1000uF/6.3V capacitor is relatively small. I am quite sure you can fit a pair in there somewhere. My source for capacitors this size is old 486/Pentium mainboards. And yes, you can always solder capacitors in parallel (their capacitance adds up). Check Rod Elliot's website (yes, I am a fan of his too!), he has some comments on this issue.

Amp question.

Reply #56 –
Quote from: antz on
...
Capacitors (electrolytic) tend to fail as open or short circuit or just by a reduction in value.
...

I finally found a good paper on the Web about capacitors, with an analysis of electrolytic capacitor failure.
You can read it here:Parts 4 and 5 are the most relevant imho. This is by Panasonic, they should know since they have been manufacturing capacitors for quite a while now.
There are also a few notes about the use of electrolytic capacitors for audio applications and their influence on sound quality in part 6.

Amp question.

Reply #57 –
I've puy my cmoy to the test now, and as i feared, it didn't look pretty.

Have a look, Test #1 and Test #2
0dB is 2V RMS, at -10dB my headphones give 104dB of volume. It's interesting how 16 or 32 ohms doesn't make much of a difference in terms of distortion.
I want flat lines for distortion, and more backup grunt. Time for a second project, but what design to choose...
Veni Vidi Vorbis.

Amp question.

Reply #58 –
Quote from: HbG on
...
Time for a second project, but what design to choose...

For simplicity and low cost:
TDA1308 (8 pin DIP) (60c) + 4 x 3.9k 1/8W metal film resistors (10c) + 3 x 100uF 10V capacitors (30c), total = 1 euro
Just for fun, try to find the TDA1308 in the following picture:

Amp question.

Reply #59 –
My magnifying glass only served to make my LCD's pixels bigger.

-brendan

Amp question.

Reply #60 –
Here's one with grunt:

It's not portable, but Rod Elliot's Project 113 - Headphone Amplifier measured really well (and sounded fine) when I built it. It can drive phones down to 8ohm with a sufficient power supply (just be careful with the volume knob, at full power it will set fire to most phones). It's easy and cheap to build (with NE5532), and is certainly worth a try.

Amp question.

Reply #61 –
Quote from: bhoar on
My magnifying glass only served to make my LCD's pixels bigger.

-brendan


Here is a better picture of these few pixels (actually the underside of these few pixels):



My hat is off for the fine engineers at Philips who designed this.

Amp question.

Reply #62 –
Allright, you win.  I'll see if i can get that TDA1308. Is it left of C2 by any chance?

Cabbagerat: That certainly looks nice. How well do you think it'll run off 9.6v? I had also been looking into this design from circuitsonline.net:
Veni Vidi Vorbis.

Amp question.

Reply #63 –
Quote from: HbG on
Allright, you win.  I'll see if i can get that TDA1308. Is it left of C2 by any chance?
...

It's the puny little thingy to the right of C2. 

This kind of packaging is really for special applications, for example, hearing aids.

Like a lot of people in this forum will have to use a few years from now... 

BTW the design from circuitsonline.net you have reproduced above is a mono multi-watt amp; you'll need to double everything to get stereo, and just like the headphone amp from Rod Elliot (which is really just a small speaker amp), it will set your headphones/earbuds on fire quite easily.

I thought I should repeat the quite appropriate warning from cabbagerat. Personally I wouldn't use any audio gear that can seriously harm me if by accident a knob is turned the wrong way.

It seems the distortion levels you have measured for your CMoy were taken at very high output levels, where the opamp had been driven to clipping already. That would be easy to observe actually, on a oscilloscope or using your audio card to sample the output of your CMoy.

Perhaps you could make some measurements at "reasonable" output levels? I am sure it doesn't have that much distortion below clipping. Also you could determine at what dB level it starts to clip, just by watching your distortion readings.

Amp question.

Reply #64 –
Argh! To the right of C2, that's what i meant. Sorry. Amazing miniaturisation indeed.
And yes i realised the amp was mono. For circuits it seems common to only draw one channel. I'm not that green.

To keep the volume at safe levels i think i'll solder a resistor in the input before the volume pot if i decide to build the amp. Indeed i was thinking what if the knob accidentally (or maliciously) would be turned all the way up. On the other hand i would want to be able to drive high impedance headphones with it as well... perhaps a switch.

As for my cmoy, the problem is i can hear clipping at really-not-that-insane levels, and it's very annoying. At 100dB/mW my Sony EX71's aren't very efficient, which may be the explanation.
Is there any software i can use to detect the precise level at which clipping starts? Also the readings are contradictory. It doesn't voltage-clip because it reads fine with no load, but it doesn't quite current-clip either because 32 ohms actually reads a little worse than 16! I may need to do more fiddling with output levels. I'll post about that later.
Veni Vidi Vorbis.

Amp question.

Reply #65 –
Quote from: HbG on
Argh! To the right of C2, that's what i meant. Sorry. Amazing miniaturisation indeed.
And yes i realised the amp was mono. For circuits it seems common to only draw one channel. I'm not that green.

Sorry, I just meant it's a lot of hardware that you'll have to buy and assemble, x 2.
Quote
To keep the volume at safe levels i think i'll solder a resistor in the input before the volume pot if i decide to build the amp. Indeed i was thinking what if the knob accidentally (or maliciously) would be turned all the way up. On the other hand i would want to be able to drive high impedance headphones with it as well... perhaps a switch.

It's still like killing a mosquito with a cannon... but using very little gunpowder!  The bias current required to drive those big transistors into class A will still require a few watts, just to put out  a maximum 30 mW seems quite a waste.
Quote
As for my cmoy, the problem is i can hear clipping at really-not-that-insane levels, and it's very annoying.

Could you post a schematic for your current CMoy?
Quote
At 100dB/mW my Sony EX71's aren't very efficient, which may be the explanation.

I can't really say without hearing them, but it seems to me that's about average for ear buds. Anyway, at 1mW -> 100dB you can't listen to music very long without hearing damage. 1mW RMS over 16 Ohm is in the region of 150mV peak-to-peak I think -> 10mA.
http://www.abelard.org/hear/hear.htm
Quote
Is there any software i can use to detect the precise level at which clipping starts? Also the readings are contradictory. It doesn't voltage-clip because it reads fine with no load, but it doesn't quite current-clip either because 32 ohms actually reads a little worse than 16! I may need to do more fiddling with output levels. I'll post about that later.

Using Linux, you can generate a pure 1kHz sinewave tone with amplitude varying from -60dB to 0dB quite easily with Audacity. And analyze distortion using baudline (also doubles as an oscilloscope, so you can *see* what's happening).
I assume the same can be done with some Windows packages.

The advantage of using a real oscilloscope is that you can detect oscillations in the 30-100MHz region, which modern opamps seem to be prone to... You just can't see these oscillations with a sampled sound card input.

Going back to the TDA1308, since I just noticed you live in The Hague, you could ask Philips for free samples of the chip and even a development board (such as the one shown above). I don't know what their policy regarding samples is, but it's worth a try.

Amp question.

Reply #66 –
I did more measurements. I used the frequency generator bundled with my kX drivers and a freeware package called VisualAnalyser 8.14. At 1000hz it is distortion free at exactly -10dB, at 50Hz -10,7dB, this should correspond to 170mV RMS but i can't measure AC very well at low voltages with my multimeter so i couldn't check. With my phones that gives me about 1.8mW. Should be enough but it doesn't explain why i hear clipping in music at less volume. At 50Hz it uses 60mA of current according to my meter (i'm not sure how it handles varying currents), idle current with the phones plugged in is just 3mA.

There's nothing wrong with using a cannon to kill a mosquito, when you can afford the cannon.  The idle power for that transistor amp design is bad news though, i wanted to run it off battery power.

A schematic.. i can post that:


I made the following adaptations:
IC1 is a TSA922
R1 100k
R3 4.7k (low gain)
R5 50
Capacitors are MKP

And the power supply:

Except with 220uF capacitors instead because i hadn't seen this circuit yet when i built it. I added them later, without them the clipping may have been a little worse but i didn't do precise measurements. I do have a 470uF cap across the battery, and no power LED.

Another strange thing is this: the clipped waveform at 50hz looks like there is some protection circuitry kicking in. It reaches the top and then the voltage drops faster than it should, about halfway down the level is normal again. It looks as if a little chunk is taken out of the top right of each wave. I think this is some protection circuitry in the TLE2526 which can only take 20mA of current and 170mV into 16 Ohms is 10.64mA, times two. As i feared.

Edit: I think that in this design i'd need really large caps to offload the rail splitter, right?
Veni Vidi Vorbis.

Amp question.

Reply #67 –
Wow, circuit diagrams from Crocclips... that brings back memories. Awesome program for quick mockups of designs 

Quote from: HbG on
Another strange thing is this: the clipped waveform at 50hz looks like there is some protection circuitry kicking in. It reaches the top and then the voltage drops faster than it should, about halfway down the level is normal again. It looks as if a little chunk is taken out of the top right of each wave.
Protection is exactly what is happening. Driver chips like this protect themselves very well against shorts, and are amazingly robust - but you trade sensible clipping behaviour for that robustness.

Amp question.

Reply #68 –
Quote from: cabbagerat on
Protection is exactly what is happening. Driver chips like this protect themselves very well against shorts, and are amazingly robust - but you trade sensible clipping behaviour for that robustness.


Thanks for the confirmation. Here's what it looks like btw:


It is only from about 300hz down that this is visible, above that it looks just like normal clipping.
What's really wierd is that it is actually worse at 32 ohms, though only very slightly.
Veni Vidi Vorbis.

Amp question.

Reply #69 –
Quote from: cabbagerat on
...
Protection is exactly what is happening. Driver chips like this protect themselves very well against shorts, and are amazingly robust - but you trade sensible clipping behaviour for that robustness.

Me thinks it's the voltage splitter (rated at +/- 20mA) that's clipping, not the op-amp (rated at +/- 80mA)...

That doesn't look at all like an amp clipping, it looks rather like a capacitor discharging...

Amp question.

Reply #70 –
Nice screenshot!
Suggested changes:
1) Get rid of the voltage splitter and the 2 x 220uF capacitors. Your new ground becomes negative of battery.
2) Leave the 470uF cap across the 9V battery.
3) Add a 470uF in series with R5 (positive terminal to R5, negative to speaker). One per channel, of course.
4) Lower R5 to 10 Ohm.
5) Connect return wire from speaker to negative of battery.
6) Decouple opamp with a 100nF ceramic cap directly across its supply pins (4 and 7).
7) Change R3 back to 1K Ohm (that increases gain to 11).

re-test after changes under exactly same conditions.

It would be a lot easier with a protoboard...

Amp question.

Reply #71 –
Thanks for the suggestions, i'll get a protoboard and try them. It'll mean building a new amp too, the soldering on this one is already too much of a mess.

Should the cap be inside or outside the feedback loop and won't it affect the sound? I built this thing for bass response, i wouldn't want to lose that. Also since the cap blocks DC could i get rid of C1?
Veni Vidi Vorbis.

Amp question.

Reply #72 –
Quote from: HbG on
...
Should the cap be inside or outside the feedback loop and won't it affect the sound?

The capacitor should be outside the feedback loop. It will affect the sound in two ways:
1) Capacitors are slightly non-linear. You will get some 2nd-order harmonic distortion (which you can measure using your software), my guesstimate is around 0.1~0.2%, which is not audible. I am interested in your results once you have tested it.
2) The RC constitutes a high-pass filter, but it's difficult to calculate the -3dB frequency because the impedance of your ear buds will affect the response curve.
Quote
I built this thing for bass response, i wouldn't want to lose that.

If you measure the frequency response and find out that it rolls off at 50Hz, you can double the capacitance of the output capacitor, or double the value of the resistor -> the frequency response will now roll off at 25Hz. Pretty simple really.
Quote
Also since the cap blocks DC could i get rid of C1?

C1 is there to block DC from whatever audio source you are using. I recommend keeping it. Quoting from C. Moy's article:
Quote
The high-pass filter C1-R2 at the input blocks DC current and has a corner frequency of about 15Hz. Substituting a 1uF capacitor will lower the corner frequency to 1.5Hz

Very honestly I don't think your earbuds have much output at 20Hz, so I would just leave C1R2 alone.

Amp question.

Reply #73 –
Actually because they are canalphones my buds go down to 6Hz. I googled the formula and 470uF with 16 ohms gives me a 21Hz cutoff freq, which should be acceptable. That is assuming the actual impedance at those frequencies is 16 to begin with..

Will any electrolytic cap do or do i need some special type?
Veni Vidi Vorbis.

Amp question.

Reply #74 –
Quote from: HbG on
Actually because they are canalphones my buds go down to 6Hz. I googled the formula and 470uF with 16 ohms gives me a 21Hz cutoff freq, which should be acceptable. That is assuming the actual impedance at those frequencies is 16 to begin with..

Will any electrolytic cap do or do i need some special type?

What can you hear at 6 Hz?  Just kidding... 

The capacitor can be any electrolytic type with a 10V rating or higher. You can of course also use a 1000uF/10V electrolytic or 2x470uF in parallel, since capacitors with this voltage rating are small. There are even some miniature ones but no need to worry about that, the normal ones will fit easily in your case.