Thursday, June 22, 2017

Rebooting the Krell

Every owner of a FPB 300, FPB 600, or subsequent Krell amplifier should know this.  I don't see it in my owners manual, so either you have to figure it out (as I have done...though I think I read about it somewhere but can't find the original discussion now) or, contact Krell (using form on website) with a problem, and they may tell you to try this first.

Manually flipping the breaker at the back of the amp, and/or unplugging from the wall for awhile, is one way which can fix certain kinds of problems, notably the problem I had been having for the last 3 weeks in which I blamed virtually every other component first (I should have thought of testing the Krell back during the first panicky week, but since it just came back from service, since I know it can output well in excess of the 5 amps required to blow Acoustat fuses, and I didn't want to believe the Krell was at fault, perhaps only input/output cables, AC power, the speakers, etc).  The amplifier was shutting down after less and less time, first time it shut down after several hours, then almost an hour, and finally it took a mere 5 minutes to get to shutdown.  (It was not blowing the breaker in back, that is serious stuff and you must contact Krell then.  My FPB 300 was shutting down with the power light still illuminated, and no other lights, and the manual says that means either inadequate AC power, or shorted speaker or cables.)

Finally, just as I was about to contact Krell, since it had finally seemed I had proven the amplifier was at fault because the problem stayed in the same amplifier channel even after I changed both the inputs and outputs to the ones normally used by the other channel, using a now dedicated 20a line, it occurred to me I should try the unplugging trick.  So I did, on Wednesday night, 3 weeks after I started having problems.

Most recently it seemed that using the right channel of the Krell would cause a shutdown within 5 minutes.  But after unplugging the Krell for a few hours, I played 90 minutes full tilt without a problem.

This is pretty good evidence this makes a difference, perhaps less that I have finally fixed the problem, but I really hope so of course.

I think what happens is that the Krell while still AC powered saves information about the bias voltages required to reach different plateau levels.  This enables the plateau thing to work as advertised.  But if the Krell is subjected to a catastrophic situation, one in which fuses blow, for example, it can store bad info, and this bad info leads to future misbehavior.

Now, this could simply be a bug in the original plateau bias system (which is managed by a chip I've seen on the schematic, I haven't looked closely but I usually call this "the Krell computer").  Maybe later amps fixed this "bug."  But it's also a kind of feature.  If a user is doing bad stuff, such as running against AC power limits as I think I may have been sometimes doing (until I moved the subs to a different circuit this month) the company wants you to contact them so they can advise you to do things differently.

Also, take this as a correction or addition to the stream of posts I was making this month while trying to track down this problem.  After I had given up trying to tweak my way around the shutdowns, I "determined" (incompletely and falsely, apparently) that the Acoustat speakers were at fault.  I had seem the Acoustat MK-121-C interface fail before--that happened to me in 2009.  That time, instead of the amplifier shutting down without any sign of distress, it blew fuses.  It blew fuses more and more often, and the problem quickly went away when I replaced that interface unit with a New Old Stock replacement interface unit I purchased on eBay for $75 more than I had paid for the pair of speakers with two interfaces.  It seemed like a bit of a ripoff then (it wasn't at all, such things are unobtanium and only a distant dream now, I was simply amazing lucky, as if the heaven of Acoustat lovers wanted to keep me in the fold) but I thought it was well worth it, now that I had fallen in love with Acoustats.

Let me add, though there has been some great fear at times, mostly this month has been lots and lots of fun....trying stuff, seeing what different things do, and being forced to play more music because I had to do more tests.  (Sometimes, quite often actually, I don't play music because I'm feeling down and I don't want to take the trouble to figure out what to play--that's the hardest part for me.  But when I've got to figure something out, I have GOT to play something.)

My biggest fear actually was that I'd be stuck with a somewhat malfunctioning new $2500 dac which the factory wouldn't fix (because they aren't responsible for complex SPDIF setups or something) and couldn't unload either because it was customized and when I told any potential buyer of the problem I was having (that is always my policy btw, full disclosure) they'd move on too.

So when it finally appeared that the DAC was not the source of the problem (because problem went away when I disconnected the right speaker) I was inclined to immediately go with the new idea that the speaker was at fault, even though I did not check to see whether it was the right speaker OR the right amplifier channel that was doing this.  I should have tried the right speaker on the left channel.  An obvious thing to do to a thinking audiophile, but I wasn't thinking too clearly.  Also, I strongly didn't want to blame the amp because though it has been fully repairable by the factory so far, it's also costly and huge amount of effort to pack and ship the amplifier back.

I finally did do the required test this week, and found the problem always remained in the right amplifier channel even driven by the left DAC channel and playing the left speaker.  For a day, I had been planning my message to Krell.  But when writing a friend about the problem, the reboot idea came back to me.  I did try that last year before contacting Krell, but it didn't help last year, the amplifier truly needed the Capacitor Service it got, it was 18 years old after all, and came back operating better than it ever had before in my 9 year experience (leading me to believe now it was actually a bit off when I bought it).

This time, rebooting fixed the Krell, I hope, and problems will disappear for long enough to consider it a perfect repair.

Update: Further investigation has revealed the Krell does indeed have an 8Mhz microprocessor in each channel to control the plateau bias, monitor DC offset and other bad situations, and prevent using the amplifier in different world region than intended.  This is discussed at DIYAudio.

The microcontroller in FPB 300/600 is a 68HC711E9.  The generic form of this now discontinued microcontroller has 512 bytes of ram, 512 bytes of EEPROM, and 12kbytes of EPROM, along with 8 channels of 8 bit A/D conversion.  The specific 711E9 version seems to more like 10 channels of A/D conversion, or so it seems from the schematic.   They connect to various sensors and circuits in the amplifier.  Other than this microcontroller, in some ways the FPB is simpler than other amplifiers, because it does not have specific protection circuits of various kinds like typical amplifiers do--where often most of the complexity is.  The microcontroller handles all the protection issues, and can shut down the amplifier with various combinations of lights lit.  I wonder if this microcontroller isn't a descendent of the Motorola 68000 series originally used in Mac and Amiga and the earliest Sun computers.  The first Amigas also ran at  8 Mhz.

I don't know this as a fact, but it has seemed like that 512 bytes (or more) of ram inside the microcontroller could get corrupted by certain situations, requiring the power cycling I have described.

One thing I might have done to corrupt it was this.  I was changing cables and for some reason I had to get underneath the power cord.  So I flipped the back breaker and temporarily removed and replaced the power cord, then flipped the breaker back on.  This would not have been a problem EXCEPT for one thing: the amplifier was already hot when I did this, and I went back to full power operation immediately after restarting.  If the amplifier does some sort of calibration when AC power is started (as I am guessing) if you reset the breaker when the amplifier is warm it might get the bias calibration wrong.  Krell advises using the soft touch power button to go into standby instead, but of course you can't do that if you are removing the power cord.

From now on, if for any reason I remove power or flip the breaker on the back, I will wait until the amp cools before restarting.  If this is the actual cause of the shutdown issues I was seeing, it's a real hoot because I had so many other ideas.

Now I remember a similar situation occurred just before I put the amplifer in my air conditioned storage building in 2011 for 2.5 years.  That time I was trying to reduce "clinking" by moving the amplifier to the left side of the room.  So I flipped breaker, unplugged and replugged in the new location a couple minutes later.  Then I turned on amplifier while still hot, and it shut down within a few minutes.  Fearing a major problem, I put the amp away.

When I took the amp out of storage I had the nerve to try it again.  And then, except for the right channel always getting much hotter (a problem fixed during capacitor service in February 2017) it ran fine for almost a year.

No comments:

Post a Comment