OK. I am flummoxed at the moment. One of my two amplifiers, both AA-1800's, apparently has an issue no one has been able to find. Occasionally, there will be heavy crackling in the right channel when the amplifier is turned on. Sometimes, it will go away, but I've discovered the hard way if it does not, the channel will fail, and fail hard. This was an eBay purchase sold with the right channel not working. So whatever is wrong has been there and not found by several techs.
Other times, the amplifier powers up perfectly and will perform flawlessly.
The latest failure took out 2 TO-3 transistors and two additional transistors in the corresponding circuit. One TO-3 was a wrong part, and one of the others was also incorrect. First hint is this part of the circuit has failed before and repaired incorrectly. Correct parts were ordered and installed and it worked perfectly.... for a time.
Due to age, I decided to replace all 16 TO-3's in the amplifier with a drop in replacement. Many of the old transistors were leaky, so it was a worthwhile thing to do.
I've also replaced all of the electrolytic caps in both channels as well as the larger carbon composite resistors,
The problem still persists. Occasionally, the right channel makes the dreaded cracking sound, and FYI, I have to replace two 10 drivers with damaged voice coils due to the latest failure.
So the million dollar question. What the heck am I overlooking?
Other times, the amplifier powers up perfectly and will perform flawlessly.
The latest failure took out 2 TO-3 transistors and two additional transistors in the corresponding circuit. One TO-3 was a wrong part, and one of the others was also incorrect. First hint is this part of the circuit has failed before and repaired incorrectly. Correct parts were ordered and installed and it worked perfectly.... for a time.
Due to age, I decided to replace all 16 TO-3's in the amplifier with a drop in replacement. Many of the old transistors were leaky, so it was a worthwhile thing to do.
I've also replaced all of the electrolytic caps in both channels as well as the larger carbon composite resistors,
The problem still persists. Occasionally, the right channel makes the dreaded cracking sound, and FYI, I have to replace two 10 drivers with damaged voice coils due to the latest failure.
So the million dollar question. What the heck am I overlooking?
I see the lesser AA-1600 (125 wpc vs. 250 wpc, schematic here) uses diode string biasing (3 + another 2). Sometimes these bias diodes go intermittent, eventually resulting in catastrophic failure of the output stage. An appropriate number of small-signal diodes wired in parallel can be used to investigate this.
That one doesn't have a cascode in the input stage either and may be running input transistors on the edge, another thing to investigate.
That one doesn't have a cascode in the input stage either and may be running input transistors on the edge, another thing to investigate.
I would vote bad solder joint too.
Or a trimmer gone faulty.
Or a bad connector.
Another possibility is oscillation. You could verify that with a scope on the output.
I never connect an amplifier direct to a speaker.
I always put a DC protection module in between.
Must have saved me a fortune in speakers over the years.
Or a trimmer gone faulty.
Or a bad connector.
Another possibility is oscillation. You could verify that with a scope on the output.
I never connect an amplifier direct to a speaker.
I always put a DC protection module in between.
Must have saved me a fortune in speakers over the years.
Next on my list of "what to try" is to replace the two germanium diodes. Someone modified the circuit and mounted them to the back of the board.
Pro-built, amateur built doesn't matter. A bad solder joint can happen on a pro-built unit, like my PV-1.3k that had the output transistors replaced so many times, the DC power was disconnected by the "tech" and a sticker "DO NOT USE CHANNEL A" installed on the back panel. The root cause was a dodgy solder joint on the feedback pin of the input op ampsocket. Probably made at the factory in 98 and not found until I traced it in 2013. Only let go when the unit was warm. Use a hair dryer to speed up "warm". The bad "joint" can be a bad weld inside a resistor or transistor, so don't get too hidebound. Look at the evidence on the pins.
OEM number parts don't matter much. They won't be identical however they are labled. Those 197? processes are abandoned. New parts should be higher rated than the old ones. MJ21193/94 is as tough a TO3 part as a solid rubber tire, $4 each at mouser/newark, and will replace most any TO3 silicon BJT. MJ15015/16 for low wattage 1980's parts is sometimes cheaper, but not always.
If the unit oscillates due to fast parts used (4 mhz Ft) instead of slow ones (200 khz) there are workarounds involving $1.50/channel in disk capacitors. You detect oscillation with a scope, or a $25 VOM on 20 VAC scale with a .047 capacitor in series with the negative probe. No music, should be no AC voltage. You can prove AC voltage is ultrasonic by changing to a 400 pf capacitor. If it is still there, it was high frequency, not music.
The spreader diode stack moved to the output transistor heat sink is an improvement, not a defect, IMHO. I'm going by the AA-1515 schematic which is available online.
Instead of ge (germanium) diodes which are dodgy and were unreliable always, use schottky diodes. Make sure back voltage rating is higher than the application, which is not hard these days with 100 v diodes available, but costing $1 typically.
To find these ittermittant problems, put a back to back >3300 uf >80 vdc capacitor in series with each speaker. It sounds funny at high frequencies and low volumes, but will protect your speaker when the amp whangs into DC out mode. You need the speakers connected to hear the big pop when the amp goes defect mode. Listen to the radio through the amp until the channel fails. I have $4 car radio speakers from the charity resale shop I use for test, not my good ones. One is 4 ohms, hasn't blown up my 8 ohm rated amp yet over coupla hour test times.
For test put the amp on a light bulb box. This is a grounded metal box with a circuit breaker, a power cord, an Ac socket, and a 100 W old fashioned incandescent bulb inside, connected hot to hot plug to socket. Use a real ceiling socket with screws and crimp connectors, not some flying wire solder to steel base disaster. Plug the amp in the light bulb box instead of the breaker strip. When the amp is fine the light bulb will be out. When it is wrong, producing DC it will light the bulb instead of blowing up TO3 transistors. I have an extra conduit hole in the back of my box so I can see the light shining. It is an old NEMA 12 controls box.
Warning, new "incandescent" bulbs sold in stores now,(I bought sylvania at family$) have a diode in series now, they are not suitable. ) If you absoluetly can't find one, use a coffee cup immersion heater. Keeping the water in the cup is a nuisance, though, I've kicked my light bulb box many times while changing records.
Stupid DC voltages on the part instead of on the the board side will finger bad solder joints. My problem was so touchy the problem would come & go when I probed the leg of the part involved with the meter. So, that was the bad joint. Reheat the joint, a little added solder, no more problem. $800 pro quality amp bought for $60 and $90 parts. (most new parts came from PCAT ATX supplies, especially all the blown diodes and .22 uf disk caps. $80 was the output transistors and drivers).
Best of luck.
OEM number parts don't matter much. They won't be identical however they are labled. Those 197? processes are abandoned. New parts should be higher rated than the old ones. MJ21193/94 is as tough a TO3 part as a solid rubber tire, $4 each at mouser/newark, and will replace most any TO3 silicon BJT. MJ15015/16 for low wattage 1980's parts is sometimes cheaper, but not always.
If the unit oscillates due to fast parts used (4 mhz Ft) instead of slow ones (200 khz) there are workarounds involving $1.50/channel in disk capacitors. You detect oscillation with a scope, or a $25 VOM on 20 VAC scale with a .047 capacitor in series with the negative probe. No music, should be no AC voltage. You can prove AC voltage is ultrasonic by changing to a 400 pf capacitor. If it is still there, it was high frequency, not music.
The spreader diode stack moved to the output transistor heat sink is an improvement, not a defect, IMHO. I'm going by the AA-1515 schematic which is available online.
Instead of ge (germanium) diodes which are dodgy and were unreliable always, use schottky diodes. Make sure back voltage rating is higher than the application, which is not hard these days with 100 v diodes available, but costing $1 typically.
To find these ittermittant problems, put a back to back >3300 uf >80 vdc capacitor in series with each speaker. It sounds funny at high frequencies and low volumes, but will protect your speaker when the amp whangs into DC out mode. You need the speakers connected to hear the big pop when the amp goes defect mode. Listen to the radio through the amp until the channel fails. I have $4 car radio speakers from the charity resale shop I use for test, not my good ones. One is 4 ohms, hasn't blown up my 8 ohm rated amp yet over coupla hour test times.
For test put the amp on a light bulb box. This is a grounded metal box with a circuit breaker, a power cord, an Ac socket, and a 100 W old fashioned incandescent bulb inside, connected hot to hot plug to socket. Use a real ceiling socket with screws and crimp connectors, not some flying wire solder to steel base disaster. Plug the amp in the light bulb box instead of the breaker strip. When the amp is fine the light bulb will be out. When it is wrong, producing DC it will light the bulb instead of blowing up TO3 transistors. I have an extra conduit hole in the back of my box so I can see the light shining. It is an old NEMA 12 controls box.
Warning, new "incandescent" bulbs sold in stores now,(I bought sylvania at family$) have a diode in series now, they are not suitable. ) If you absoluetly can't find one, use a coffee cup immersion heater. Keeping the water in the cup is a nuisance, though, I've kicked my light bulb box many times while changing records.
Stupid DC voltages on the part instead of on the the board side will finger bad solder joints. My problem was so touchy the problem would come & go when I probed the leg of the part involved with the meter. So, that was the bad joint. Reheat the joint, a little added solder, no more problem. $800 pro quality amp bought for $60 and $90 parts. (most new parts came from PCAT ATX supplies, especially all the blown diodes and .22 uf disk caps. $80 was the output transistors and drivers).
Best of luck.
Last edited:
http://i56.photobucket.com/albums/g196/dkleitsch/dualdiffcascode.gif
The AA1800 was a basic Leach (double-barreled) amplifier design.
I would look at all the small plastic transistors, and for gold-leaded 2N3440/5415 parts.
Tin in solder migrates into the gold and makes for intermittent connections. Remove (wick) the old gold contaminated solder, and replace with a good solder (with some silver content if possible).
Copperweld leads on 5% carbon film lead resistors fail when the copper striker coat lets go of the steel lead. Replace the resistors. Some film caps have copperweld leads too, replace.
The AA1800 was a basic Leach (double-barreled) amplifier design.
I would look at all the small plastic transistors, and for gold-leaded 2N3440/5415 parts.
Tin in solder migrates into the gold and makes for intermittent connections. Remove (wick) the old gold contaminated solder, and replace with a good solder (with some silver content if possible).
Copperweld leads on 5% carbon film lead resistors fail when the copper striker coat lets go of the steel lead. Replace the resistors. Some film caps have copperweld leads too, replace.
- Status
- Not open for further replies.