http://www.ozvalveamps.org/repairs/gibson.html | Created: 14/01/06 | Last update:
A G-70 rides again
Date: 6th Aug'02
Maree is a keyboard player and singer in a local Reggae band and she brought in a “combo” guitar amplifier to repair for use on-stage and recording with her classic Yamaha DX-7 synth.
“I think one of the valves is gone” she said, in what would turn out to be something on an understatement.
I looked at the heatsinks on the back; then inside, not a valve or socket to be seen.
When I told her it was actually fully solid-state she was skeptical “are you sure?”
I finally took the chassis out to show her “look ma, no valves” but she didn't seem to be convinced.
Originally it was a Gibson G-70 made under contract by Chicago Musical Industries (CMI) but this one had done a lot of hard yards and had been seriously modified at some point.
In retrospect, perhaps I should have been more cautious when I spotted the modifications.
At first glance several of the controls had frozen, the wooden case was coming apart and the covering peeling off.
Inside was an even bigger mess.
It had stopped working some years before and had been stored in a shed where it suffered the ravages of humidity and rodents.
It was a very typical “combo” (combination) setup with the amplifier chassis suspended upside down by four screws in the top of the slotted-back case above two 12-inch speakers, and a spring-line reverb “tank” in a leatherette bag screwed to the bottom of the open back case.
Some repairers see a lot of salt-air corrosion. Being in the bush I see a lot of critters such as mud wasps, spiders, and rodents.
This amp innards were covered with old webs of harmless tiny “money spiders” who seem to instantly colonise the insides of everything, so first I hit it with the vacuume cleaner and brushes.
With the chassis cleaned up and on the bench it had stories to tell.
The most obvious thing was that the output and driver transistors had originally been thermotabs mounted internally on the steel chassis as a heatsink. It beats me why any manufacturers still think they can get away with this in arduous guitar service, least of all a name like Gibson, when few Hi-Fi amp makers would risk it.
Two beefy heatsinks had been added to the chassis rear and the output pair had been changed from an unknown thermopad to TO-3 2N3055's, with TIP31/32's used as drivers and VAS (Voltage Amplifer Stage).
The emitter resistors were intact, but burned to a crisp - not a hopeful sign.
The pots were newish but a motly collection including stereo controls were none are needed.
Splat marks showed where the mains switch had worked loose at some time and touched the chassis.
The cable clamp was missing, and the mains lead had been replaced with a perished rubber (not very) insulated horror.
At least half the pots were obviously duff in some way so I decided to replace all nine. Whoever had fitted them had used pots with metric splines while the collet clamp knobs were for the old standard 1/4-inch shaft. The difference had been made up with a binding of electrical tape, another reason to change the lot.
I ordered all the pots, but when they turned up they didn't match the catalogue description, having imperial D-shafts that were 1mm undersize! They went back and the correct pots were obtained from a different supplier. This kind of supplier carelessness make me foam at the mouth because of the disproportionate and unrecoverable time and effort it costs.
In the meantime the mains wiring was cleaned up and the PCB removed to replace the burnt 2 watt emitter resistors with 5 watt 0r22's.
Close visual inspection turned up two broken components, a 1k resistor in a protection circuit, and the 0.1uF cap in the Zobel(*) VHF stability network on the output.
Neither would have stopped the amp working, so they could have been broken for years. More serious was a lifting track under the VAS transistor that was ready to fail.
With these components replaced and the pots finally fitted and static tests showing no suspect transistors in the output stage, it was time to gingerly switch on for the “smoke test”.
Joy! It was a goer.
But my joy was short-lived.
As I advance the input level I could see I was getting asymmetric clipping at high level, driving the signal centerline up on the CRO.
Normally I don't have a lot of trouble with speakers, but as soon as I saw the trace centerline moving I started to wonder about the health of the speakers, if they had been exposed to excessive DC.
I cursed myself for getting in so deep without first checking the speakers, and sure enough both had open voice coils, so it was pretty much a write-off when it came in, but I was in far too deep to back out now.
In some ways this was better than losing only one of a pair where you may have to find a match to a 16 ohm, high stiffness, high power driver - a rare animal these days. As luck would have it I did have a new pair available, so in they went (but only after Murphy made sure they had different mounting holes).
Back in the amp it was now imperative to find the cause of the clipping ten volts short of the negative rail, that is if I didn't want to boof these new speakers as well.
In short order I progressively changed the bootstrap capacitor (and incidentally a couple of other sorry looking bypass caps), the lower 2N3055, then both drivers for trusty BD139/140's.
After this flurry of fruitless activity I made myself a coffee and rolled a smoke (which, like choice of loud music, is one of the advantages of a solo workshop) and sat down to have a serious meditate on the chassis.
I have developed an idea over years of faultfinding (ranging from mine safety to Intensive Care) that the real problem is you, not the equipment. The fault simply is what it is, just lying there waiting for you to recognise it. It isn't hiding, or being difficult, it's just that you haven't looked at it yet in the right way to see it.
This often bruses young egos, but ego and assumption are the enemy of effective faultfinding - “just concentrate on the facts, Watson”.
So, I have been assuming (from the burnt emitter resistors and dead speakers) that I'm chasing a partly-failed transistor. On reflection that doesn't sound very likely.
I am also assuming from the workman-like modifications that they are okay.
But are they?
The soft clipping suggests that something is “zenering out”, that is non-destructively being driven past its breakdown voltage.
Apart from the obviously changed 3055's, are the others the original design transistors I start wondering?
I had already replaced both drivers against possible breakdown. Now looking at the pre-driver VAS, its collector at full cutoff swing is subjected to both supply rails, 70-odd volts. Hummm.
Out with the databooks to check the voltage rating of a TIP31 - 60 volts. What is so baffling one moment is so obvious the next. Replacing it with a higher voltage BD140 finally ended the chase, giving full swing and hard clipping at both rails.
So this one was modified and improved in respect of the transistor heatsinking, only to have another serious design flaw introduced in the process.
No wonder the poor thing wound up banished to the shed.
My weak justification for this sidetrack is that I had noticed that the output stage was thermally unstable (using a millivoltmeter clipped across the emitter resistors to monitor the idle current) because the compensation diodes were on the PCB, nowhere near the output pair.
With the heatsinks already off to replace the output and driver transistors I took the opportunity to mount another BD140 on each so its E-B diode could be used to sense temperature instead.
Counter-intuitively these track temperature better through the top than the pad, so work better in this duty mounted upside-down. Note that this is only a physical change and that the transistor junction functions electronically just as the diode except it is now thermally coupled to the thing it was supposed to sense in the first place. A mod but not to the effective circuit.
In later tests the bias current was utterly stable from cold to maximum power point temperature.
Next thing to look at were the effects.
“Vibrato” (actually Tremolo, an amplitude modulation that produces a subjective pitch shift) is applied using a variable low frequency oscillator driving an “opto-coupler” comprising a lamp and Light Dependant Resistor combination.
Despite the fragility of the (original) tiny filament lamp this all still worked fine.
Next was the reverb.
These generally are a unit about 2 ft (600mm) long containing two (or more) springs of different type to give different delays. These are suspended between taut-wire magnetic driver and pickup units which may look identical but are very different impedance. Sometimes they have a spring arrestor lever fitted to stop the unit being damaged during transport.
Some rodent had found the reverb pouch a comfortable place to set up home and have a family. It was absolutely stuffed with straw and “other” material, so at first I didn't give it much chance, but after applying the vac and brushes it came up almost as new.
Like most of these units it was stamped “input” and “output” so I hooked it up on the bench and (another broken wire in the amp aside) it worked fine.
After a bit of glue and strap-clamping the wooden case was again solid and it was time for it all to go back together.
On test everything was fine until I got to the reverb.
It was very weak and there was a physically sensitive intermittant oscillation that was almost supersonic.
The reverb line came out again and a bit of prodding showed that one of the RCA connectors was loose to the line chassis. To avoid an earth loop the unit is only grounded on the drive side, the pickup side screen requiring continuity of ground through the unit. This ground was one of the socket retaining rivets, and this had worked loose.
Making this good cured the instability but the reverb was still weak.
I had refitted the reverb line as I had found it, but investigation discovered two holes hidden on the far side of the pouch that aligned with the RCA sockets, very obviously for the connection leads. In other words the line had some time previously been replaced back-to-front.
This corrected, everything worked okay. That weekend I had the pleasure of mixing a Reggae show where Maree used the amp and it performed perfectly all night.
As a keyboard player myself I normally wouldn't even consider using an open-backed guitar amp because I have a heavy left hand style that requires a solid bass.
But Maree mostly “skanks” above middle-C and this amp suits her lighter guitar-like style reasonably well.
Musicians tend to have a lot of trouble with electronic repairers because it is a specialised field. The gear looks superficially simple like a home stereo and their concerns often sound absurd in the workshop.
But after you have done a few shows you start to understand the daily equipment and sound struggle that most musicians live with. Lugging, setting up and tearing down a truckload of gear several times a week, often in new and odd situations.
It's a mixture of art and business where deadlines are real and equipment is “mission critical” to income. Reliability is everything. Band equipment gets a hard life and must be mil-spec tough, as do any repairs.
This was a “name” amp, already modified for apparently good reason, but some older name amps are also genuine classics that must be serviced with an eye to maintaining their originality, if possible, as well as their functionality.
Old lables and chassis markings in particular can be vital to establishing authenticity and must be preserved.
Maree has decided to follow my suggestion to strip the badly worn covering and re-cover it using nurseryman's root balling cloth, a very tough and cheap alternative to speaker or boat carpet.
Then its renewal will be complete and it will literally be better than new. Something more than just a repair, and a very satisfying outcome.
Maree still calls it her “valve” amp. ;-)
(*) A Zobel network is a resistor and capacitor in series, “almost invariably 10 ohms and 0.1uF” from the amplifier half-rail (inboard/before any stability inductor) to ground. The action of this network is generally poorly understood, but is actually to prevent local HF/VHF instability in the output stage. These traditional values have been confirmed to be around optimum.
Ref: “The Audio Power Interface”, Doug Self, Electronics World, Sept 1997.