BGW 750E Hi Current Fault

[nevetslab]

Every so often my past catches up with me, being one of the product design engineers for BGW Systems. Brian Wachner’s eldest son Jeffrey called me out of the blue a couple weeks ago, and had a pair of 750E’s, a 350A and a PS2 Power Amp that all need service. I was still in the middle of getting the two ATI Paragon Power Supplies restored, but had him drive up to the shop to drop them off.

Yesterday, having finished up with the last two Ampeg SVT-CL amps from AuntieM, as well as their two Roland JC120’s, an Eden WT800 and a Fender Deluxe, I had a brief lull in the work load, so I moved the two 750E amps up onto the check-out bench. One I knew would not power up (FP switch is a 20A Magnetic Breaker, and it wouldn’t allow powering), the other had both RED CLIP LED’s lit.

The 750E was built with NEC 4-lead 200W Power xstrs that Brian had purchased some 10 million pairs of that last production use for at least 10 years. But, when they were gone, they were gone. No second source. I knew BGW’s best chief engineer Duke Aguiar had kept a few hundred pairs, and had been selling them on ebay….no idea if he still had any left. Duke had designed this amp around those NEC parts, also did the PCB layout, while I assisted with the input PCB’s and the harness design, allowing modules to be connected outside the chassis for service.

So, I opened up the first one, extracted the power amps to see what the status was on it’s power xstrs. Both channels NEC parts all measured fine…no shorts. Opened the second 750E chassis up, and found those also were fine, so that was a relief.





That did leave me with the question of what’s causing the circuit breaker to fire? I plugged it into the power analyzer, variac fully CCW, and with the breaker on, began advancing the AC mains. Drew high current right away. I unplugged the buss caps from the 25A bridge, and tried again. It powered up just fine. Turned back down, connected the (-) buss caps, and tried again. Those charged up ok. Turned back down, disconnected the (-) caps, connected the (+) buss caps and tried those. Drew current immediately. Just to be thorough, I disconnected the AC leads from the bridge, as well as the (+) caps, and checked the bridge. It measured fine, so one or both of the (+) buss caps are bad.



These are 2” dia 7/8” spaced Screw Terminal computer grade caps, 10,000uF/100V parts, 5-5/8” tall. I checked to see what Mouser had in 100V 2” dia computer grade caps. They had some 9000uF/100V for around $52 ea. Serial number on the amp is 85E prefix, so 1985 date code. Buss caps probably 1983 or later.

What did I have on hand? I have a number of 1st generation Compact Monitor Systems TRIAMP’s, all using 20,000uF/100V buss caps, same form factor. I had one here in the shop, plus a GTC (also Compact series), and a 2-channel Prototype built with the motherboard for the TRIAMP, set up as a 2-ch input Mono Subwoofer amp, built in a non-painted TRIAMP chassis, and it had a pair of 20,000uF/100V buss caps. I hadn’t powered it up since the 9/11 attack in NYC. It too was sitting here on the floor.



I moved it up to the bench and opened it up. It had a pair of GTC power amp modules in it, one being a spare, the other one connected. I connected it to the power analyzer/variac, and slowly ran it up in steps, to see if it would power up, monitoring the DC buss voltage along with the AC Mains current. I had reached 90VAC, with the supplies at +/- 65VDC. As I advanced the variac, I suddenly saw the mains current increase, and then heard a loud crack, as I turned the variac down to 0VAC, then looked inside the chassis to see if there was a trail of smoke. Discharged what was left on the buss caps, disconnected the GTC power amp module, expecting to find a shorted device. Didn’t find anything on it, so moved the chassis to the test bench for closer inspection. I found a piece of shrapnel from a power device….7815….looked at the LM7815T regulator on the power supply heat sink, and saw it had blown off the device. The TRIAMP has a +/- 40V regulated power supply to run the Tweeter Amp, fed off of the +/- 85VDC supplies, as well as a pair of NPN/PNP Pass Xstrs to pre-regulate to +/- 24VDC zener supply to run the 7815T/7915T bipolar supplies to run the front end/crossover & filters for the TRIAMP. The 24V zener was shorted, as was the 2N3773 NPN Pass Xstr, so no doubt the 1000uF/35V buss cap had failed. Only had one of those fail over the years that TRIAMP had run stage monitors for Compact Monitor Systems. Still running after 30 years!

But, while I would have to pull the motherboard out to repair that damage, I did get high enough on the buss caps to feel confident in using that pair to replace what had failed in the 750E.



I then pulled the hold-down clamp off of the pair of 20,000uF/100V buss caps, then did the same in the 750E chassis. Removed the buss bars joining the pairs of caps and the long ground buss bar, and removed the (+) pair of caps, moved one of the (-) caps over and rotated it, and moved in the pair of 20,000uF/100V buss caps, and formed a new 30,000uF/100V buss cap array for this 750E. Put the hold-down clamp back on, cable-tied the wiring back into place, then connected the chassis to the power analyzer/variac, DMM to the buss caps, and slowly ran it up to verify I have a functional chassis again.

Powered up just fine. No modules connected, so tally lights were lit accordingly. That’s tomorrow’s project.

[J M Fahey]

Only one word: IMPRESSIVE!!!!!!

[nevetslab]

I connected the pair of amplifier modules to this 1985 chassis, and brought the AC Mains up slowly thru the power analyzer on the variac, while watching the bias voltage on each amp module and AC Mains current. All was nominal.. DC output level was under +/- 2mV. I powered it down, discharged the supplies, then powered it up the normal way with 120VAC mains, and all came up normal each time. The only abnormal item I see is the Power Supply Imbalance LED is lit, though both the +/- supplies measure +/- 85.53VDC. So, I'll have to pull the display PCB apart to find the problem in that comparator circuit. As luck would have it, I had loaned all of my BGW Service Manuals to Duke Aguiar, former BGW Chief Engineer, who was expanding his on-line document service. He's been behind in completing that task. All I have in my database is what I was able to download, where I'm finding the display board listed in the 750D/E manuals containing that from the 750F/G, which is not the same, nor even the interface connections.

So, hoping Duke has that correct display document on hand that can be sent to me. I then connected signal source to the 1/4" input jacks....where the inside jacks are on the secondary side of the plug-in transformer sockets, and the outside jacks are in parallel with the loop-thru Male/Female XLR connectors. This module is strapped differently than the same module in the other 750E chassis. As a result, I only get signal thru the amp channels using XLR connectors.

I powered up the other chassis with it's amp modules, bias was correct, and everything came to life just fine. DC output level a little higher, less than 10mVDC on both channels. Initially, powering that amp up had the RED Clip LED's lit, but no longer the case. This 1986 generation amp powers up fine, and its' 1/4" input jacks are working correctly. So, I'll have to get to the bottom of that in the morning. So, progress made, along with a couple mysteries to solve.

[nevetslab]

In checking my email this morning, I found Duke had sent me a BGW 750D/E manual, but, upon opening it, I found it was the same one from HiFi Engine that's floating around on the internet, which does NOT have the correct 750E Display PCB. It has the one from the 750G...a different circuit. Nor did it or the other schematics he sent have the details on the 9008-0001 Xfmr Input module, which has the jumpers for assigning the parallel pair of input jacks to be either in parallel with the XLR's or on the output side of the xfmr sockets which feed the front panel detented pots. I looked thru my collection of blank PCB's I had laid out at BGW, but didn't find that board in the collection. So, back to the normal procedure of figuring it out again.

[nevetslab]

I removed the Transformer Input module on this 1985 Vintage 750E, so I could revise the TRS Connections to the XLR connectors. Stock mode had TRS Tip connected to XLR Pin 3, and TRS Ring to XLR Pin 2. The Polarity jumpers set at REV route XLR Pin 2 to Xfmr Pin 7, and XLR Pin 3 to Xfmr Pin 8. Neither amp had the TRS jacks properly phased, and only one had the XLR to input xfmr properly phased So, I needed to remove the TRS jumpers to be replaced with insulated jumpers making TRS Tip connect to XLR Pin 2, and TRS Ring connect to XLR Pin 3 for Pin 2/TRS Tip (+).



Here, I removed and replaced the jumpers with insulated crossed jumpers correcting the TRS to XLR polarity, as well as the Polarity jumpers to the input transformers on this 1985 model.



I moved the 1986 model to the bench to correct it's input board to match that revision made to the 1985 model.



Now, both amps are identical as far as input signal polarity of the TRS jacks and XLR connectors, and both pass signal just fine. Next issue is dealing with the Display PCB on the 1985 model, which is showing the RED LED on the Power Supply Imbalance light. Both supplies are equal in magnitude, so there must be something up with the comparator circuit.

Jensen made input transformers for these amps (750B/C, 750D/E, 750F/G, 250D/E, 250F/G, 350A, GTA, GTB), as well as Triad, and later Pacific Transformer (who made most of the later power transformers for these models).

After speaking with Duke on that large purchase of NEC power xstrs, it was 110 thousand NPNs and 124 thousand PNP's that we had purchased, not 10 million pairs. Big difference, but they did last a long time. There were years of producing quasi-comp amps with those NEC parts, which over time left us sitting on a huge pile of PNP's and the NPN's were getting used up. Changed to PNP-based qusi-comp design versions of the existing products to even out the balance, before we moved onto other available power xstrs.

NOTE:
I've no idea what happened with inserting the photos on the post. It got totally corrupted, and took three iterations to get it to this state, but, all the blank space below this body of the text I can't get rid of.

[nevetslab]

I took the power amp modules back out of the 1985 model, as this was the one that had the Power Supply Imbalance LED lit. It wasn't until I already had the display removed and checked continuity of the +/- supply lines from the buss caps to the connector. That's when I found I wasn't getting the +85VDC line at the connector. Looked at the terminal connections on the two buss caps and saw the small ring lug was missing, and found that wire laying on the chassis floor, it having fallen out off the terminal screw and missed it. So, that wasn't a display defect. Restoring that took care of that LED being lit.



I put that amp back together, turned it around front facing out, and moved the other amp to the bench stacked on top of it, connected short T/S patch cords in to daisy-chain the four inputs together and drove it with pink noise.




I did find the Line Voltage Monitor LED's on this amp were not showing the same threshold as that of the 1986 model. On the 1986 model, Low Line turned on at 105VAC, and High Line turned on at 129VAC. The 1985 Model, Low Line turned on at 112VAC, and I didn't turn the AC Mains up past 132VAC with the modules connected, but after removing the connections to the bridge rectifier, I found I got it turning on at 140VAC. That sparked recollection of the complaints we had with this model, as the circuit was not built around precision components, always having a predictable threshold. We discontinued that display feature on the 750G for that reason. And, if we had sought UL approval, we no doubt would have had trouble with 120VAC presence on that display board and the clearances req'd.

So, there's a difference between these two amps in that regard. I forgot about that until now. And, not having the circuit diagram yet for the 750E Display, I didn't have any luck staring at the PCB with all the parts loaded onto it. I could tweak it to be accurate with the documents, but lacking that, I put the display back together, even finding one keps nut kissing a nearby circuit trace, and fetched a tiny lock washer and hex nut, removing that potential short.



I checked both amps for residual noise, using an XLR Male plug with a 604 ohm termination resistor across pins 2 and 3, and plugged that into each channel as I measured the noise level on the Amber 3501A analyzer. The 1985 model measured 50uV RMS on both channels 20kHz BW, while the 1986 model measured 100uV RMS on both channels, same BW. I haven't yet checked output level and distortion. I'll look at that later.

[nevetslab]

I sent an email to the owner of these amps, along with a link to this thread. He wasn't aware of the rear input module being the Transformer Input version, which we had standardized on when the amp went into production. In 1985, I had laid out the PCB for the 2-Ch Active Balanced input module to retrofit onto the connector plate that his 9007-0001 Xfmr Input PCB mounted, so we could eventually make the change to using the Active Balanced Input module as standard, and the Transformer Version as an option. I don't recall when (or if) we made that change.



I couldn't recall how I had powered the Active Balance Amp assembly until I looked at the supporting schematic for it. The Power Xfmr only had a 12VAC winding to run the display and turn-on voltage to the power amp modules to turn on/off the constant current source in the front end of the amp. So, I had powered the Active Input module for this and it's next versions off of the +/- 85V Supplies thru 2k/5W power resistors to 16V zeners on the module.

BGW-9008-0002 Active Input Module.pdf

Looking at that pen and ink schematic I made in doing this document, I see at the time BGW was still one of many in the industry using Pin 3 as (+), Pin 2 as (-), before the industry finally adopted the IEC Microphone polarity standard Pin 2 (+) for all line level and audio signal polarities. Hence 'Normal Polarity' and 'Reverse Polarity' markings on the drawing. I think it was Ampex who was the main force behind Pin 3 (+) being the polarity standard. We had adopted that for years before finally changing.

I looked briefly last night for any Jensen 0900-9050 and 0900-9150 Bridging Input Transformers we had them make for our amps. Also Triad made a similar 10k Bridging Plugin Xfmr 0900-9250, and later had that produced by Pacific Transformer, where Randy Neverka had learned his chops from the masters over at Peerless Transformer (Altec Lansing). I did see offerings for all of the above in the past few years, but didn't spot anything currently offered. Only way to get those active on the input modules I set up for identical strapping is to remove the jumpers bypassing input and output of the transformer sockets, so you could plug in any of those input transformers. The Jensen xfmrs had much better shielding than the Triad and Pacific Transformer units....at least there's a Toroidal Power Transformer in the chassis, but, the input transformers would be sitting right above them. There is mounting provisions for a hold-down clamp, or a shielded container, though we never did stamp one out as I recall.

I checked the distortion characteristics briefly yesterday on both my Amber 3501a Audio Analyzer, as well as on the Bruel & Kjaer 2010/1902 Analyzer, which will allow plotting individual harmonics (2nd, 3rd, 4th and 5th harmonics in Harmonic Distortion mode, as well as Difference Frequency Distortion and Intermodulation Distortion)...plotting those harmonics over frequency. I have the XY plotter 2308 that will drive the system, but, all my fiber tip pens have dried up, and, to my surprise, even the sealed ink containers evaporated to source the older plotting pen that you have to push thru the sealed container and use the cleaning tool to prime the ink flow, as well as unclogging the ink flow. Messy, but no fiber tips to dry out. I had picked up some fine old hypodermic syringes and needles for use with fine machine oil as well as refilling ink cartridges. Now I have to find that tool box in storage where those hypo's are stored. You used to be able to buy those on ebay years ago.

I never did work out an interface to digitize the DC Drive for the Frequency Axis as well as the Vertical axis to record plots generated by the B & K gear. Still living in the land of pre-printed K & E graph paper and ink pens for the results.

At any rate, distortion levels were respectable, in the 70-90dB below fundamentals at low thru high listening levels. Unless you're using really inefficient speakers, who is listening at 200-500 Watts output levels anyhow? That's headroom power.10-50W is considerably loud. Crossover distortion in the 1W level area was the worst....below -60dB, but not down in the mud, either. If I get some time, I might run some plots on the spectrum analyzer and take photos of that.

[nevetslab]

I spent some time with the B & K 2010/1902 Distortion Analyzer, got readings at 1V, 3V, 10V & 30V out at 1khz and 5khz, with images of the readings at fundamental, 2nd, 3rd, 4th and 5th harmonics, along with images of the waveforms on the scope (fundamental and THD distortion waveform. I've shown the amp at 10V output driving 8 ohms (12.5W Output), typical listening level).



Then, looking at the Bruel & Kjaer 2010/1902 Analyzer showing 1kHz Fundamental @ 10V Output/12.5W 8 ohms, then 2nd, 3rd, 4th and 5th harmonics. The meter full scale on 2nd thru 5th harmonics has a 50dB range, FS -50dB. As the Harmonic control is rotated to 2nd (-78dB), 3rd (-84dB), 4th (-88dB) and 5th (-87dB), the Analyzer retunes to 2kHz, 3kHz, 4kHz and 5kHz



This would be much easier to read on the plotter, so there's be swept distortion over a 3 decade range. Best I can do at the moment, for lack of working pens on hand. Or, of course, on an Audio Precision System. Never had the funds to spring for one.