Teac AG15D Tuner-Amplifier

 I bought this rather nice equipment ages ago to replace my old Tandberg tuner amplifier and I've been very pleased with it since, until fairly recently when the volume control knob stopped working so we had to use the remote control, then very recently every so often a loud crack from the speakers would deafen you. This loud crack would daily set the amplifier back into standby, but it could then be put back on again until the next occurrence of the loud crack... then the other night the loud crack changed to a quiet crack but every few minutes instead of once a night. I diagnosed dry joints and decided it was time to disconnect everything and fix the problem. Disconnecting stuff is a horrible job as the tangle of connections have materialised over many years and then removing the lid from the amplifier reveals it's not going to be an easy job.


 Once you are familiar with an equipment it's a lot easier to work on it and of this Teac was a perfect example. Firstly it looked problematic to remove the main circuit board, but after removing 46 screws I had the thing out on the bench. At this point I can advise that the rear panel needs not to be detached. I initially removed countless screws and did this, but later when still looking for the problem, discovered the rear panel can be left attached to the main board.


 Inspecting the solder side of the main board revealed lots of cracked solder joints. A clue to their locations were several areas of discolouration due to local heating. The worst dry joints were the connections to three voltage regulators fastened to a vertically mounted heatsink. As there were three small electrolytic capacitors inaccessibly close the the heatsink, I removed the heatsink complete with the three regulators then fitted two new 1uF and a new 10uF which are connected to the regulator outputs. I then checked the other electrolytics in the area where heat had discoloured the board and found that the dreaded securing compound that I used to encounter when I repaired VCRs had been used to secure the larger capacitors. This compond gradually turns into a brown brittle substance and during this process liberates a highly corrosive by-product that eats copper and tin. Fortunately only a single wire strap (which was anyway duplicated) had been partly corroded. Before I'd realised it was that nasty securing compound I'd detached a 4700uF capacitor that I thought had leaked. In fact the capacitor seemed to have leaked but possibly due to the effects of the securing compound. As it looked pretty horrible I cleaned the circuit board and fitted a new capacitor. I also removed the 5 volt smoothing capacitor and fitted three 2200uF capacitors wired in parallel as I did not have a small 6800uF handy. After carefully refitting the heatsink assembly and resoldering around 30 cracked solder joints I reassembled everything and refitted the multitude of cables.

Incidentally.. that nasty gluey stuff that turned brittle and corrosive was once used in a complex chip used in a Fujitsu hard drive. All examples eventually failed in slightly different ways causing their owners lots of grief. A large computer firm called "Tiny Computers" who had used these hard drives went bust. I think eventually the chip manufacturer bit the dust and most hard drives got a bad name until memories faded.


 Switching on I was alarmed to hear a loud hum from the speakers followed by the amplifier switching itself off. I must have shorted something in the resolder exercise or perhaps even fitted a capacitor back to front. All the cables had to be disconnected again and I settled down to identify the problem. This time I'd realised that the rear panel can be left attached so the zillions of screws remained in place. First... had I fitted a capacitor wrongly? No all were fine. Had I shorted something when resoldering? I didn't think so but decided to buzz out the connections as that loud hum I'd heard might have been a power rail shorting, but all were OK.

Below, immediately left of that central heatsink are three defunct capacitors, looking just as innocent as the scores of similar, but good ones.


 Just in case an audio power transistor was bad I removed the output assembly. This is secured by three screws underneath and two screws either side of the assembly. At each end is a small board that needs to be loosened then the whole amplifier board can be lifted out. There's a long row of power transistors which appear to be complementary NPN/PNP pairs with a bias transistor between them. A check with my multi-meter showed all six circuits were perfectly symmetrical so no problems there. I also tested the standby circuit board in case it's capacitors were bad and were somehow turning off the equipment but no problems found here either.



 Here are the suspect capacitors. Left to right 1uF open circuit, 1uF open circuit, 10uF which had an ESR greater than 20 ohms and measured about 2uF, 1000uF which had a good ESR but was only 800uF so was on its way out. The other two are 4700uF and 6800uF which I swapped but then found the horrible gunge was remnanants of a securing and highly corrosive glue. This would eventually eat the terminals.
 As I progressed, occasionally powering the main board whilst carefully balanced outside the chassis and hearing the thing turning off after a few seconds, I came across something rather odd. The 5 volt ground connection at the regulator pin wasn't connected to the main copper ground circuit. Both the adjacent 7812 12 volt regulators were grounded but not the centre pin of the 7805. I traced the circuit and found the 7805 connections were indeed marked GND but these were megohms away from chassis. Neither was the main power supply which is around +/- 40 volts connected to chassis. As this is also marked GND I was puzzled. Maybe that last loud crack had finally blown a hole in one of the printed circuit tracks? That would explain why it had turned into a quiet crack?

 I found a repair manual on the Net and studied the track layout for the main board circuit. This showed the 5 volt ground was indeed not connected to the 12 volt regulator ground connectons. But why were both circuits labelled GND?

In studying the circuit I noticed a selection of small lower power circuits that seemed to be sensing various voltages and connected to the circuit that turns off the amplifier. Maybe one of the dozen capacitors in this area had failed and was making the protect relay operate? This can happen if a delay circuit of say 5 seconds using say a 100uF capacitor whose value had dramatically dropped or whose ESR had risen sky high was not allowing time for a voltage to establish? However all the capacitors, although not perfect were certainly good enough for purpose. I then decided to look with a magnifier for a break in the ground circuit perhaps caused by a component failure. I found two interesting things. These were metal tabs, through which two, otherwise discrete and isolated, ground circuits connect to the chassis. That was the answer to the problem. When I'd detached the main circuit board for the second time I'd noticed one loose securing screw. This (because of Sod's Law) connected the main 40-0-40 volt supply ground to chassis (hence the loud hum when first installing the repaired amplifier). The second tab, which had been tight, connected the 5 volt ground to chassis. This method of grounding is probably used to minimise circulating ground currents which might result in hum or modulation of sensitive amplifier circuitry. Below... the critical securing screws which must be tight.




 I refitted the main board and plugged in all the cables, turned on the amp and waited for the usual few seconds... but this time the amplifier stayed on.

I suspect the initial fault was just a selection of dry joints at the 5 volt and 12 volt regulators. Before I reassemble everything I'll look into that duff volume control... possibly more dry joints? And of course in these digital days the control is not a simple potentiometer but a pulse generator whose pulses are counted by a microprocessor. No wonder it's failed (mechanical plus elelectronic = eventual failure) and of course switch cleaner would be ueless. I decided to leave the thing in the end because it uses a special rotary switch.


 I opened up the old Teac once again and looked at the main circuit board. For some reason the amplifier had been cutting out and we relied on a good thump on its case to resurrect the output. I found two probable reasons. One was dry joints at relay coil connections and the second was a slip-up I'd made when fitting a cooling fan. I'd used the one of the three regulator outputs to provide me with the fan supply voltage but inadvertently increased the regulator current making it even hotter. I rewired the fan to the main supply voltage. There were two.. plus 47 and the minus 47 volts. I decided to check the fan to see if it would run more slowly than at its rated 12 volts and decided on 4 volts, at which it drew 110mA. I fitted a large resistor of 390 ohms (47V-4V/110mA= 391 ohms) and this worked with exactly 4 volts across the fan terminals with fairly unobtrusive noise. Having put back the AG-15D into service I tackled the AG-D200. I'd been using this for a week or so and found it had several faults. One was failure to drive several of its speakers, and after some period it would make sharp cracking noises. Also, I found the whole HDMI system was inoperative (which is its main feature).

Below, the newer Teac doesn't use too many straightforward plugs and sockets so must be dismantled rather differently!


 Below a view of the main culprit for cracking noises...



 What you can see above is a circuit board (mounted in front of the power output transistors on seven plugs). To remove this board needs some extensive levering but it then pops off for scrutiny. There are seven identical-looking preamp and driver transistor sets. You'll notice the pre-amp transistors have heatsinks (those vertical silvery-coloured things). Using a magnifier you can see that at least 50% of the solder joints have circular marks which are cracks in the solder. I started by resoldering all the transistor legs which were by far the worst affected but to make a lasting repair I'll need to resolder nearly all the solder joints.

Why is this board so bad? My guess is the solder is almost certainly an early lead-free type whose composition is not up to its job. I can also see that the amount of solder on many joints is rather lacking in quantity. Interestingly all seven pcb connector soldering is fine, but where even the slightest heat is present the more heat the more cracking so it's clear the fatigue strength of the solder is very poor. I noticed lots of red markings on this and other boards and I wonder if this represents evidence of QA examinations during manufacture?



 Here's a small area affected by solder cracking. The shiny joints, I've already resoldered as these were the worst, but you can see that nearly all are affected to some extent. In this area the audio signals are extremely small and readily susceptible to extraneous noise.

I counted 7 amplifier circuits each carrying 100 joints. I resoldered 700 joints on this board and around 100 on the main board plus a couple of others. I realised I wasn't the first because of copious numbers of red ink marks on many of the solder blobs.


I managed to get all the bits put back together except a small metal bracket which could of course belong to virtually anything I've had in for repair. I plugged in and switched on and everything seemed OK but I'll wait until after Christmas before refitting into my Hi-Fi system as there are far too many leads to move around. In the meantime my old AG15D is doing a sterling job.



 Return to Reception