Maker Pro
Maker Pro

Help diagnosing audio amp distortion

T

Terry Pinnell

Jan 1, 1970
0
I was hoping to finish off this little stereo amplifier today. It's a
very simple design based on two LM380s, and I've placed it in the same
case as my existing multi-purpose bench unit (which contains various
preamps and mono power amps). It shares its 13.3V regulated DC supply
but nothing else apart from accommodation.
http://www.terrypin.dial.pipex.com/Images/StereoAmp1.gif

But I get nasty distortion (on both channels), whenever input
amplitude (at jack socket) exceeds the very modest level of about
700mV.

http://www.terrypin.dial.pipex.com/Images/DistortionMusic.gif
http://www.terrypin.dial.pipex.com/Images/DistortionSine.gif

I've done a fair bit of checking but so far can't see whatever the
obvious cause is!

BTW, why is the 'polarity' of the distortion pulses different for
those two cases of sine versus music please?
 
F

Fred Bloggs

Jan 1, 1970
0
Terry said:
I was hoping to finish off this little stereo amplifier today. It's a
very simple design based on two LM380s, and I've placed it in the same
case as my existing multi-purpose bench unit (which contains various
preamps and mono power amps). It shares its 13.3V regulated DC supply
but nothing else apart from accommodation.
http://www.terrypin.dial.pipex.com/Images/StereoAmp1.gif

But I get nasty distortion (on both channels), whenever input
amplitude (at jack socket) exceeds the very modest level of about
700mV.

http://www.terrypin.dial.pipex.com/Images/DistortionMusic.gif
http://www.terrypin.dial.pipex.com/Images/DistortionSine.gif

I've done a fair bit of checking but so far can't see whatever the
obvious cause is!

BTW, why is the 'polarity' of the distortion pulses different for
those two cases of sine versus music please?

The obvious cause is the Absolute Maximum Rating on Input is 0.5V.
 
A

Andrew VK3BFA

Jan 1, 1970
0
Terry Pinnell said:
I was hoping to finish off this little stereo amplifier today. It's a
very simple design based on two LM380s, and I've placed it in the same
case as my existing multi-purpose bench unit (which contains various
preamps and mono power amps). It shares its 13.3V regulated DC supply
but nothing else apart from accommodation.
http://www.terrypin.dial.pipex.com/Images/StereoAmp1.gif

But I get nasty distortion (on both channels), whenever input
amplitude (at jack socket) exceeds the very modest level of about
700mV.

http://www.terrypin.dial.pipex.com/Images/DistortionMusic.gif
http://www.terrypin.dial.pipex.com/Images/DistortionSine.gif

I've done a fair bit of checking but so far can't see whatever the
obvious cause is!

BTW, why is the 'polarity' of the distortion pulses different for
those two cases of sine versus music please?


OK, will probably be shot down for saying something really stupid, but
why do you have the input to bothe the inverting and non inverting
input pins - see page 3 datasheet. Just a thought....
Andrew VK3BFA
 
T

Terry Pinnell

Jan 1, 1970
0
OK, will probably be shot down for saying something really stupid, but
why do you have the input to bothe the inverting and non inverting
input pins - see page 3 datasheet. Just a thought....
Andrew VK3BFA

Same valid question raised by Roger in
alt.binaries.schematics.electronic

I cribbed the circuit virtually unchanged from a mono
amp in Graham Bishop's 'Audio Circuits & Projects', shown here:
http://www.terrypin.dial.pipex.com/Images/StereoAmp-Bishop.gif
 
F

Fred Bloggs

Jan 1, 1970
0
Terry said:
Thanks - I guess that *is* obvious! Where did you find that spec
please? I can't see it in this datasheet:
http://212.57.231.17/datasheets/69/OTPWONOVSO.pdf

It also seems there's some doubt about the actual circuit, which I
sourced from:
http://www.terrypin.dial.pipex.com/Images/StereoAmp-Bishop.gif

I can't pull your datasheet in, but the National data sheet makes this
very clear. You might also check National AN-69 which is a short
covering the fundamentals. Your circuit seems right to me, it is using
what National calls common-mode input signal attenuation, which is a
means of adjusting the output level without attenuating the input signal
level by adjusting the input differential signal presented to the fixed
gain amplifier. But you still have to keep that single-ended amplitude
something like 250mV maximum. At a fixed gain of 50 this would be 25Vpp
at the output- you need to ask yourself what the heck are you doing with
all that single-ended amplitude when you attenuate the differential by a
factor of 5 for maximum output.
 
R

Ralph & Diane Barone

Jan 1, 1970
0
Terry Pinnell said:
I was hoping to finish off this little stereo amplifier today. It's a
very simple design based on two LM380s, and I've placed it in the same
case as my existing multi-purpose bench unit (which contains various
preamps and mono power amps). It shares its 13.3V regulated DC supply
but nothing else apart from accommodation.
http://www.terrypin.dial.pipex.com/Images/StereoAmp1.gif

But I get nasty distortion (on both channels), whenever input
amplitude (at jack socket) exceeds the very modest level of about
700mV.

http://www.terrypin.dial.pipex.com/Images/DistortionMusic.gif
http://www.terrypin.dial.pipex.com/Images/DistortionSine.gif

I've done a fair bit of checking but so far can't see whatever the
obvious cause is!

BTW, why is the 'polarity' of the distortion pulses different for
those two cases of sine versus music please?

As others have noted, the absolute max input voltage for the LM380 is +/-
500 mV (page 3 of the National data sheet. Note that whenever you apply
any negative voltage to the chip, you are applying a signal that's outside
the power supply rails. The National engineers even give themselves a pat
on the back in the data sheet by saying "A unique input stage allows ground
referenced input signals".

I would change your input pot to a more traditional design, with the input
connected across the pot, the wiper connected to pin 2, and pin 6 left
floating.
 
J

James Meyer

Jan 1, 1970
0
As others have noted, the absolute max input voltage for the LM380 is +/-
500 mV (page 3 of the National data sheet. Note that whenever you apply
any negative voltage to the chip, you are applying a signal that's outside
the power supply rails. The National engineers even give themselves a pat
on the back in the data sheet by saying "A unique input stage allows ground
referenced input signals".

The input circuit looks to me like it's current driven and not voltage
driven like a common op-amp. In that case, you aren't really supplying an input
voltage, but an input current.
I would change your input pot to a more traditional design, with the input
connected across the pot, the wiper connected to pin 2, and pin 6 left
floating.

All the application circuits show just the opposite. Are you saying
that you know how the circuit operates better than the circuit designer?

Jim
 
R

Ralph & Diane Barone

Jan 1, 1970
0
James Meyer said:
The input circuit looks to me like it's current driven and not voltage
driven like a common op-amp. In that case, you aren't really supplying an input
voltage, but an input current.


All the application circuits show just the opposite. Are you saying
that you know how the circuit operates better than the circuit designer?

Jim

Are you asking if I know more than the designer of the LM380, or the person
who came up with the original design at the start of this thread? The
answer to the first is "hell no" and the answer to the second is "who
knows?".

Since the LM380 spec sheet claims that the maximum input voltage is +/- 0.5
V, I would assume that all the example schematics in the LM380 datasheet
using that style of volume control do not have input voltages that exceed a
half a volt. Since the original poster's system obviously does not have
that limitation, it would seem prudent to attenuate the signal going into
the input stage, rather than rely on cancellation in the diff amp. I
suspect that the style of volume control shown in the data sheet may give
the resultant circuit slightly better PSRR or SNR at low volumes, but I
suspect that sacrificing a bit of noise is worth it to get rid of that
awful input stage saturation behaviour.
 
F

Fred Bloggs

Jan 1, 1970
0
James said:
All the application circuits show just the opposite. Are you saying
that you know how the circuit operates better than the circuit designer?

Jim

All the applications show no such thing. The traditional single-ended
pot degrades S/N with attenuation ratio, the differential scheme shown
on the data sheet does not. Ordinarily this would not be a problem, but
when you use big pots like the 2.5M in the data sheet, then it is a
problem. The calculation breaks down like so:

Please view in a fixed-width font such as Courier.

Rs
Vs o---/\/\--+
|
| e
/ |/
Rp \ <----+--|
(pot)/ k | |\
\ _ / |
| Rb |
| / |
| \ | Rb=150K
| | |
+-------+------+----+
|
---
///



Pot noise is sqrt(4KTRpB), eg Rp=2.5M-> Vn=20uV in 10KHz BW


For fixed attenuation k, input signal and noise become:


Rb||(kRp)
Vsi=Vs x -------------------------
Rb||(kRp) + (1-k)Rp + Rs


Rb /--------
Vni=Vn x ------------------------- , Vn= \|(4KTRpB)
kRp((1-k)Rp+Rs)
Rb + ---------------
Rp + Rs


kRp((1-k)Rp+Rs)
Rb||(kRp) x (Rb + ---------------)
Vsi Vs Rp + Rs
---= -- x -------------------------------------
Vni Vn (Rb||(kRp) + (1-k)Rp + Rs) x Rb



- or -


Vsi Vs Rp Vs
--- = k x -- x ------- = k x -- for Rs<<Rp
Vni Vn (Rs+Rp) Vn



WHEW!!!! attenuation dependent S/N


Now compare to:

Vsi Vs Rp
--- = -- x -- for the differential rheostat
Vni Vn Rb


left as exercise for the reader........
 
T

Terry Pinnell

Jan 1, 1970
0
James Meyer said:
The input circuit looks to me like it's current driven and not voltage
driven like a common op-amp. In that case, you aren't really supplying an input
voltage, but an input current.


All the application circuits show just the opposite. Are you saying
that you know how the circuit operates better than the circuit designer?

Jim: As a side issue, I don't have that post to which you replied,
dated Sun, 01 Feb 2004 16:52:04 GMT, from [email protected] (Ralph &
Diane Barone). Nor can I find it in Google News.

Fred: Before seeing your reassurance about the original circuit, I
changed it to the more conventional one shown here:
http://www.terrypin.dial.pipex.com/Images/StereoAmp2.gif

Actually, it's one of several 'conventional' configurations I found.
One option appears to be whether to leave pin 2 unconnected or
grounded. I'm unclear what the respective diffrences are. I also
assume, although haven't tried this, that you can symmetrically
interchange pins 2 and 6. (Or would the non-inverting version increase
risk of unwanted feedback, even at this low gain level?)

Anyway, I'm in final stages of testing this revised circuit at
present. Initially it was awful, with hf oscillation. Seem to have
solved that by
- adding a second 0.1uF bypass close to second channel amp (one was
apparently fine in the original circuit)
- adding overall 47uF supply bypass electrolytic (hadn't got around to
doing that in first circuit, and, apart from the problem described, it
seemed to tolerate it).

BTW, note that as I didn't have a ganged 2.2M pot, I'm using 200k. The
revised circuit source (Marston, 'Linear IC Pocket Book', pg. 154) has
a 47k, but I assume the difference here is not significant.

Also, I'm still not sure that the original circuit's misbehaviour was
due to the input of 500mV being exceeded. As mentioned, I was getting
that noise when the signal *at the jack* (i.e ahead of pot) exceeded
about 700mV, regardless of pot setting.
 
J

John Woodgate

Jan 1, 1970
0
I read in sci.electronics.design that Terry Pinnell <terrypinDELETE@dial
Also, I'm still not sure that the original circuit's misbehaviour was
due to the input of 500mV being exceeded. As mentioned, I was getting
that noise when the signal *at the jack* (i.e ahead of pot) exceeded
about 700mV, regardless of pot setting.

If you look at the internal circuit as given on the data sheet or AN-69,
where Figure 9 shows the differential gain control), you can see that,
wherever the pot is set, large negative input voltages drive an input
bases (one or both, depending on the pot setting) below the
collector(s). Not good. Large positive excursions of the -ve input could
do nasty things to the complementary current mirrors in series with the
Darlington input device. Also not good. That AMR of +/- 0.5 V MEANS IT!
 
F

Fred Bloggs

Jan 1, 1970
0
Terry said:
Fred: Before seeing your reassurance about the original circuit, I
changed it to the more conventional one shown here:
http://www.terrypin.dial.pipex.com/Images/StereoAmp2.gif

Actually, it's one of several 'conventional' configurations I found.
One option appears to be whether to leave pin 2 unconnected or
grounded. I'm unclear what the respective diffrences are. I also
assume, although haven't tried this, that you can symmetrically
interchange pins 2 and 6. (Or would the non-inverting version increase
risk of unwanted feedback, even at this low gain level?)

The two inputs are pulled to GND by 150K resistors and you leave
open/GND/ GND through series R etc depending upon source impedance and
criticality of output offset. Output offset is not critical in your
application, so shorting the input like you have done is good.
Anyway, I'm in final stages of testing this revised circuit at
present. Initially it was awful, with hf oscillation. Seem to have
solved that by
- adding a second 0.1uF bypass close to second channel amp (one was
apparently fine in the original circuit)
- adding overall 47uF supply bypass electrolytic (hadn't got around to
doing that in first circuit, and, apart from the problem described, it
seemed to tolerate it).

BTW, note that as I didn't have a ganged 2.2M pot, I'm using 200k. The
revised circuit source (Marston, 'Linear IC Pocket Book', pg. 154) has
a 47k, but I assume the difference here is not significant.

Also, I'm still not sure that the original circuit's misbehaviour was
due to the input of 500mV being exceeded. As mentioned, I was getting
that noise when the signal *at the jack* (i.e ahead of pot) exceeded
about 700mV, regardless of pot setting.

Right- well your jack was piped directly into the first input in
parallel with the series combination of pot and 150K for the second
input. So you can see your pot was doing nothing about attenuating the
signal level into the first input. Also because your pot is relatively
small compared to the 150K, the most attenuation you could get at the
second input would be 150/(150+200) for a total differential of
220/(150+220)=0.6. This means your adjustment is only 0->0.6.
Looking at the application graphs, the amplifier at Vs=13V is good for a
maximum of 1.75W into 8 ohms for a THD <+10% and internal dissipation of
about 1.25W maximum, which will be about 75oC Tj rise for a modest 2.5
sq in. PC copper heat sink. Taking this as your maximum output then
means that Vout,RMS=sqrt(8 x 1.75)=3.75V,RMS for a maximum input of
75mV,RMS. Assuming you are driving this with a LINE output of 1V,RMS
implies an input attenuation of .075. If you go back to the differential
pot then the input is attenuated 0.075/0.6=0.125 before the pot which
means a circuit like this:

Please view in a fixed-width font such as Courier.



10K Vin' | \
----/\/\---+---------+----+-----------|+ \
| | | |
| / | |
| 22OK \ <--+ | LM380
/ / |
1.5K | |
Vin / +----------------|- /
\ | / |
| |
| |
| |
-----------+-----------------------+-------+-----
|
 
T

Terry Pinnell

Jan 1, 1970
0
Fred Bloggs said:
The two inputs are pulled to GND by 150K resistors and you leave
open/GND/ GND through series R etc depending upon source impedance and
criticality of output offset. Output offset is not critical in your
application, so shorting the input like you have done is good.


Right- well your jack was piped directly into the first input in
parallel with the series combination of pot and 150K for the second
input. So you can see your pot was doing nothing about attenuating the
signal level into the first input. Also because your pot is relatively
small compared to the 150K, the most attenuation you could get at the
second input would be 150/(150+200) for a total differential of
220/(150+220)=0.6. This means your adjustment is only 0->0.6.
Looking at the application graphs, the amplifier at Vs=13V is good for a
maximum of 1.75W into 8 ohms for a THD <+10% and internal dissipation of
about 1.25W maximum, which will be about 75oC Tj rise for a modest 2.5
sq in. PC copper heat sink. Taking this as your maximum output then
means that Vout,RMS=sqrt(8 x 1.75)=3.75V,RMS for a maximum input of
75mV,RMS. Assuming you are driving this with a LINE output of 1V,RMS
implies an input attenuation of .075. If you go back to the differential
pot then the input is attenuated 0.075/0.6=0.125 before the pot which
means a circuit like this:

Please view in a fixed-width font such as Courier.



10K Vin' | \
| | | |
| / | |
| 22OK \ <--+ | LM380
/ / |
1.5K | |
Vin / +----------------|- /
\ | / |
| |
| |
| |
|

John, Fred,

Thanks both. I'll study your posts. This is proving very educational,
so I'll probably also breadboard that 'differential' approach to
compare theory v actual. To be honest, I didn't properly study the
data sheet and app notes. The circuit looked so simple, I expected it
to work first time!

Meanwhile, I'm going to stick with the revised, 'conventional'
configuration, as my stripboard is beginning to resent the fair amount
of change I've already made. It's close to working and I'd like to
wrap it up fairly soon. So I'd much appreciate advice on the remaining
problems please, the first minor, the other more serious.

1. Cross-talk
Maybe this is inescapable, but with zero input on one channel I get a
modest volume from the other channel's speaker. It's not going to
worry me much in practice in this application (amplifying my Pocket
PC's headphone output so that I can listen to tracks while working in
my shed/workshop). But my curiosity is aroused as to possible causes.

2. Imbalanced performance
In short, the RH channel distorts well before the left. Swapping
speakers and LM380s made no difference. As far as I can see, both
circuits are identical. Here are a couple of illustrations:
http://www.terrypin.dial.pipex.com/Images/AmpTest4-LeftSine.gif
http://www.terrypin.dial.pipex.com/Images/AmpTest5-Imbalance.gif
 
T

Terry Pinnell

Jan 1, 1970
0
Allan Herriman said:
You seem to have left and right mislabelled (not that the amplifier
will care).

http://www.electronics2000.co.uk/data/images/jackdiag.gif

Thanks Allan. I wired input socket so that, looking at it from above,
the LH connector was on the left, and right on right, when 0V
connection (not shown) was out of the way below. That diagram you
linked therefore explains why I had the hassle of shifting my
speakers!

L R
| |
--------
| |
| |
--------------|------|-------------
--------------|------|-------------
Case front panel | |
| |
| |
 
Top