Topics

uBITX - U1 Getting Fried - possible cause #ubitx

Jim Sheldon
 

Now that I've fried 2 of the TDA2822M's in my latest uBITX once while it was just sitting there and once when I plugged a speaker in, I took a good look at things.

My take is that huge 470 uF electrolytic in the output of the TDA2822M and the speaker gets a huge surge at turn on and during any loud noise or POP in the audio chain.  It will sit there just fine with nothing connected but the instant you plug a speaker in you cause that cap to discharge and recharge again causing a huge current spike and the TDA2822M can't handle it.

The current crop of 2822's do not have a recognizable manufacturer's logo on them and I suspect they are Chinese knockoffs.  I bought a handful of them off eBay from a seller called Teufel Tronics in the US and when the original blew, I replaced it with one of these.  It ran fine until I was working on loading some test software and turned it on with no speaker plugged in.  It was sitting there fat, dumb and happy until I plugged the speaker in.  All of a sudden, BANG and the top blew off the 2822.  Fortunately I have 8 of my stock left so if it goes again, it will be an easy fix.  I got smart and put an 8 pin socket in when I changed it this last time. 

It was suggested putting a 4 or 8 ohm resistor in series with the hot lead of the speaker and I intend to do this as soon as some get here from Digi-Key in a couple days.  That may limit the surge current enough to keep it from happening again.

Jim Sheldon W0EB

Jerry Gaffke
 

I think you're right.

The output pin has a quiescent DC voltage of around half the supply voltage, or about 6v,
so the 470uF cap will have a 6v charge on it when operating.  A significant bit of energy
for the TDA2822 to provide, and it tries to do so almost instantaneously in the case of a short.
A series resistor seems prudent. 

Perhaps the 470uF cap could be reduced to 100uF.
At 500hz the 100uF would have an impedance of 1/(2*pi*500*100e-6) = 3.18 ohms,
which seems borderline for good sound quality when using an 8 ohm speaker.

I vote for the resistor.  Something like 4 or 8 ohms, 1/2 Watt.
But worth experimenting if somebody has a stash of cheap TDA2822's on hand.

Configuring the TDA2822 for Dex's  bridge configuration does away with the cap charging problem,
but now the headphone jack has to be floating.  The bridge would be ideal for an internal speaker.

Jerry


On Thu, Mar 1, 2018 at 11:47 am, Jim Sheldon wrote:
My take is that huge 470 uF electrolytic in the output of the TDA2822M and the speaker gets a huge surge at turn on and during any loud noise or POP in the audio chain. 

Tim Gorman
 

I'm not sure even a 100uf cap is needed. For communications purposes
you should be able to get by with a 47uf capacitor. A 47uf cap in
series with an 8ohm speaker should give you a low freq 3db point of
about 500hz. I realize this is a kind of a paper napkin calculation
because a speaker doesn't exhibit true resistance but it should still
be close.

You can get an 8ohm, 4" CB replacement CB speaker for about $5 on
Amazon.

I personally like the idea of a fuse in the speaker lead. That's
probably what I will do. If it is a current spike of several amps that
is blowing the 2822 then a fuse should protect it. And you don't have
to worry about wasting half your audio power in a resistor.

tim ab0wr

On Thu, 01 Mar 2018 12:16:35 -0800
"Jerry Gaffke via Groups.Io" <jgaffke=yahoo.com@groups.io> wrote:

I think you're right.

The output pin has a quiescent DC voltage of around half the supply
voltage, or about 6v, so the 470uF cap will have a 6v charge on it
when operating.  A significant bit of energy for the TDA2822 to
provide, and it tries to do so almost instantaneously in the case of
a short. A series resistor seems prudent. 

Perhaps the 470uF cap could be reduced to 100uF.
At 500hz the 100uF would have an impedance of 1/(2*pi*500*100e-6) =
3.18 ohms, which seems borderline for good sound quality when using
an 8 ohm speaker.

I vote for the resistor.  Something like 4 or 8 ohms, 1/2 Watt.
But worth experimenting if somebody has a stash of cheap TDA2822's on
hand.

Configuring the TDA2822 for Dex's  bridge configuration does away
with the cap charging problem, but now the headphone jack has to be
floating.  The bridge would be ideal for an internal speaker.

Jerry

On Thu, Mar 1, 2018 at 11:47 am, Jim Sheldon wrote:


My take is that huge 470 uF electrolytic in the output of the
TDA2822M and the speaker gets a huge surge at turn on and during
any loud noise or POP in the audio chain.

G1KQH
 

Recently I found some TDA2822's on ebay from China, I got the GQRP club to purchase a batch ( I bought 50 too ). Tony Fishpool (G4WIF) setup up a test and found although they worked well at low Volts they got hot over 7V!:


His findings are made public here on his website here: http://www.fishpool.org.uk/audioamp.htm


Food for thought!

73 Steve

GQRP 11049

http://g1kqh.blogspot.co.uk/

 

Jerry,

I changed mine to 100u in the B40s and even reduced the input 1uf as we dont need
response below 300hz. I calculated the time constant from the roll off freq and worked
out the cap.

In bridge mode you get double the power! I had originally recommended bridge to Farhan.

Raj


At 02-03-2018, you wrote:
I think you're right.

The output pin has a quiescent DC voltage of around half the supply voltage, or about 6v,
so the 470uF cap will have a 6v charge on it when operating.  A significant bit of energy
for the TDA2822 to provide, and it tries to do so almost instantaneously in the case of a short.
A series resistor seems prudent. 

Perhaps the 470uF cap could be reduced to 100uF.
At 500hz the 100uF would have an impedance of 1/(2*pi*500*100e-6) = 3.18 ohms,
which seems borderline for good sound quality when using an 8 ohm speaker.

I vote for the resistor.  Something like 4 or 8 ohms, 1/2 Watt.
But worth experimenting if somebody has a stash of cheap TDA2822's on hand.

Configuring the TDA2822 for Dex's  bridge configuration does away with the cap charging problem,
but now the headphone jack has to be floating.  The bridge would be ideal for an internal speaker.

Jerry

On Thu, Mar 1, 2018 at 11:47 am, Jim Sheldon wrote:
My take is that huge 470 uF electrolytic in the output of the TDA2822M and the speaker gets a huge surge at turn on and during any loud noise or POP in the audio chain.

Dexter N Muir
 

Yes, Raj, bridge mode gets you more, but likely more than twice! While one 'phase' is going positive, the other is going negative. W=E^2/R, double E (Voltage) gives double squared (4x) power. That was the impression I had gained from earlier discussion. More likely is the In/Out-rush current, instantaneous fubar! The heat and pyrotechnics would be after-effects. Less C for less Bass would be sane, a series R likewise - the few dB loss would hardly matter - Bitx-40's sidetone-at-the-speaker arrangement would be an indication...

73
Dex ZL2DEX

 

Dex,

Best solution would be a 6-9V regulator for the TDA. This would current limiter too..
Perhaps in uBitxV2!

Raj

At 02-03-2018, you wrote:
Yes, Raj, bridge mode gets you more, but likely more than twice! While one 'phase' is going positive, the other is going negative. W=E^2/R, double E (Voltage) gives double squared (4x) power. That was the impression I had gained from earlier discussion. More likely is the In/Out-rush current, instantaneous fubar! The heat and pyrotechnics would be after-effects. Less C for less Bass would be sane, a series R likewise - the few dB loss would hardly matter - Bitx-40's sidetone-at-the-speaker arrangement would be an indication...

73
Dex ZL2DEX

RCC WB5YYM
 

Could a person add an 8ohm resistor across the speaker jack on the NC pin? When you plug in the speaker it would open up that load and the amp would see the speaker.

Jerry Gaffke
 

That might help here, though I'm not sure we fully understand the problem yet.

This resistor would charge up the 470uF cap at power up, so plugging in a short
or headphones or whatever at a later time would not be so disruptive.
Is that what you have in mind?

Resistor could be 100 ohms, and still charge the cap plenty fast.
And if 100 ohms, could be there at all times in parallel with the headphones, 
so no need for the fancy speaker jack with the switching contacts.  

Jerry, KE7ER


On Thu, Mar 1, 2018 at 09:02 pm, RCC WB5YYM wrote:
Could a person add an 8ohm resistor across the speaker jack on the NC pin? When you plug in the speaker it would open up that load and the amp would see the speaker.

Vince Vielhaber
 

Something doesn't seem right here. At least one person said that the IC smoked as well as R75. R75 is a 2.2 Ohm resistor connected between pin 6 and ground. Pin 6 is the + input on amp #2.

Vince.

On 03/02/2018 12:20 AM, Jerry Gaffke via Groups.Io wrote:
That might help here, though I'm not sure we fully understand the
problem yet.

This resistor would charge up the 470uF cap at power up, so plugging in
a short
or headphones or whatever at a later time would not be so disruptive.
Is that what you have in mind?

Resistor could be 100 ohms, and still charge the cap plenty fast.
And if 100 ohms, could be there at all times in parallel with the
headphones,
so no need for the fancy speaker jack with the switching contacts.

Jerry, KE7ER

On Thu, Mar 1, 2018 at 09:02 pm, RCC WB5YYM wrote:

Could a person add an 8ohm resistor across the speaker jack on the
NC pin? When you plug in the speaker it would open up that load and
the amp would see the speaker.

Jerry Gaffke
 

Yup, something doesn't seem right.

From post 42723:
  I have no idea how R75 could have smoked, except maybe a catastrophic failure in the TDA2822M 
shorted that (normally) input pin to Vcc somehow.

And this from 43122:
  I fried mine when I plugged in a stereo sony noise canceling headset.
So it does not only happen with mono plugs.

And this from 43170:
   Last evening he turned his radio on and the TDA got very hot
and started smoking for no reason at all.

And this from 43211:
   Recently I found some TDA2822's on ebay from China, I got the GQRP club to purchase a batch ( I bought 50 too ). Tony Fishpool (G4WIF) setup up a test and found although they worked well at low Volts they got hot over 7V!:

Might be a few weird ones, I didn't do an exhaustive search.

A 6v regulator to feed the TDA2822 might be our best bet.

Jerry, KE7ER



On Thu, Mar 1, 2018 at 09:41 pm, Vince Vielhaber wrote:
Something doesn't seem right here. At least one person said that the IC smoked as well as R75. R75 is a 2.2 Ohm resistor connected between pin 6 and ground. Pin 6 is the + input on amp #2.

Vince Vielhaber
 

All of the uBitxes don't have the same IC. Mine doesn't look like the one Raj posted. I didn't bother taking a pic since I can't find an angle that will show the part number. I had to use a lighted loupe to see the number and even at that I had to look in a number of different angles. It just looks like a plain unmarked black IC. I can't even make out the logo - if what I was seeing was a logo.

Vince.

On 03/02/2018 01:04 AM, Jerry Gaffke via Groups.Io wrote:
Yup, something doesn't seem right.

From post 42723:
I have no idea how R75 could have smoked, except maybe a catastrophic
failure in the TDA2822M
shorted that (normally) input pin to Vcc somehow.

And this from 43122:
I fried mine when I plugged in a stereo sony noise canceling headset.
So it does not only happen with mono plugs.

And this from 43170:
Last evening he turned his radio on and the TDA got very hot
and started smoking for no reason at all.

And this from 43211:
Recently I found some TDA2822's on ebay from China, I got the GQRP
club to purchase a batch ( I bought 50 too ). Tony Fishpool (G4WIF)
setup up a test and found although they worked well at low Volts they
got hot over 7V!:

Might be a few weird ones, I didn't do an exhaustive search.

A 6v regulator to feed the TDA2822 might be our best bet.

Jerry, KE7ER



On Thu, Mar 1, 2018 at 09:41 pm, Vince Vielhaber wrote:

Something doesn't seem right here. At least one person said that the
IC smoked as well as R75. R75 is a 2.2 Ohm resistor connected
between pin 6 and ground. Pin 6 is the + input on amp #2.

ajparent1/kb1gmx
 

Its not the 470Uf cap unless it was in backward.

The problem from my analysis is that anyone running the TDA2822 on 12V is at risk. 
The parts max operating voltage is 12V!!!  Thas straight from the data sheets of three
different vendors.  Likely some can stand more but not guaranteed.  Most 12V gell batteries
freshly charged are 13.8 and even after surface charge is leveled off its 12.6.

It would be better to run it at some lower voltage as its was designed for use at 1.8V to 12V.

Allison/Kb1GMX

Arvo W0VRA
 

On Fri, Mar 2, 2018 at 10:39 am, ajparent1 wrote:
The parts max operating voltage is 12V!!!
The maximum voltage depends on the manufacturer.  The FCI version datasheet lists 16V as the maximum.  The STM datasheet lists 15V.

These WX parts are having trouble with what should be nominal voltages.

Jerry Gaffke
 

I agree, best bet is to reduce supply voltage into the TDA2822.

First reports of fried TDA2822's seemed to all be due to shorted audio output, 
as when plugging a mono plug into a stereo jack.
In series with a proper 8 ohm load, the 470uF cap should be fine,
With a short to ground, there will be a quick surge current of unspecified amps from the TDA2822 till the cap is charged.
Though if that's a failure mode, it is not the only one.  

The original ST parts (now obsolete) claimed an Absolute Max of 15v,
and also gave that as the maximum operating voltage.
This agrees with the specs for the NJ2073D and NTE7155 clones.
So running it at 12v should be legal, though nowhere else in the ST datasheet
is there mention of operation over 9 volts.  A bit of a red flag.
ST still builds the SOIC8 variant, the TDA2822D.

This UTC part gives an absolute max of 15v, but maximum operating voltage of 12v, looks legit:
  https://www.taydaelectronics.com/datasheets/A-1002.pdf

The FCI part I have on my stock uBitx has no datasheet except a clone of the ST datasheet up on one of those nasty datasheet archives:
  http://pdf1.alldatasheet.com/datasheet-pdf/view/210172/FCI/TDA2822M.html 
The FCI website itself has no datasheets:  www.cotra.com.tw

The WX parts seem to be the ones failing, have no datasheet, this is all I could find:
    http://www.wx-ic.com/Cn/Product_Detail_Images.aspx?id=105303&UserInfo_ID=216432&CorpProductImages_ID=106897

There are reports of TDA2822 clones failing at 7v:
    https://groups.io/g/BITX20/message/43331
    https://groups.io/g/BITX20/message/43211

Moral of the story:  Avoid the dodgy looking poorly documented clones.

Jerry, KE7ER



On Fri, Mar 2, 2018 at 10:39 am, ajparent1 wrote:

Its not the 470Uf cap unless it was in backward.

The problem from my analysis is that anyone running the TDA2822 on 12V is at risk. 
The parts max operating voltage is 12V!!!  Thas straight from the data sheets of three
different vendors.  Likely some can stand more but not guaranteed.  Most 12V gell batteries
freshly charged are 13.8 and even after surface charge is leveled off its 12.6.

It would be better to run it at some lower voltage as its was designed for use at 1.8V to 12V.

Paul Smith
 

Interesting developments.  Dodgy relays and IC's occur every so often due to the vagaries of batch testing but nothing will sink a new product design faster than a consistent QA problem at component level.  Some serious supplier vetting must happen to head off a possible long run issue. My 2 cents worth.

Paul ZS2OE

Paul Smith
 

How about stacking a few 1N4001s in series in the power rail to the TDA2822 ? Not ideal but could provide a couple extra volts headroom and easier to install than a regulator.

Paul ZS2OE

DrZ <zehrhj2@...>
 

I have some TDA2822 replacements ordered and will try them, but the adapter with the LM386 is working well, though the adapter will fall out of the socket if the rig is vibrated much.  If I decide to keep it that way, I may need to figure out a better adapter.

Howard K4LXY

Jerry Gaffke
 

Should work fine, as suggested in post 42723 a few days ago:
    If really paranoid, perhaps add an 8 or 9v linear regulator (or string of diodes) to reduce Vcc into the TDA2822M.

Though I'd now tend to go for 6v, especially if you have a WX part.
Just because you're paranoid doesn't mean they aren't out to get you.

Jerry


On Fri, Mar 2, 2018 at 01:13 pm, Paul Smith wrote:
How about stacking a few 1N4001s in series in the power rail to the TDA2822 ? Not ideal but could provide a couple extra volts headroom and easier to install than a regulator.

Daniel Conklin <danconklin2@...>
 

I saw the picture of that adapter somewhere, but now I can't find it. Could you link to it please? Thanks.
Dan, W2DLC