Kenwood L-07M II Issues

EchoWars

Hiding in Honduras
Sponsor
I have L-07M II #4 in pieces here on the floor, and this is the fourth one in six months that I've worked on that has oscillated itself to death

The L-07M II (not the original L-07M) is one of the fastest amps that Kenwood ever built. Kenwood rates its slew rate as 170V/µS, and says its 3db bandwidth is DC to 600KHz. Folks...that's fast. This amp was built in the days of the 'TIM' (transient intermodulation distortion) wars, and rather than raw output power, manufacturers were trying to outdo each other with speed.

While there's nothing inherently wrong with this (and a lot of things right about it), a very fast amplifier must be very stable. Kenwood has sprinkled a slew of some strange type of 'pf range' caps on the driver board of the 'M II' to ensure that stability. Problem is, these caps are apparently changing value under load and heat conditions, most likely due to age. On all three that I have running at the moment, the output Zobel filter resistors were near-charred, and driver transistors and other parts on the driver board were blown. On one (the one mentioned above), the thermal fuse (Canadian model) is open in the transformer. On another one, four of the six output transistors were blown. These are very special, high-bandwidth outputs, and cannot be sourced (a distributor in the UK says they have them...for about $200 for an amplifier set). All that can be done is to put modern output transistors into the amp, and hope it works well. It certainly isn't going to meet specifications with anything other than the originals.

Besides the danger to the output devices, there's a real danger to the transformer if the level of oscillation is high enough. Although the total current flow may not be enough to blow the fuse, the oscillation may well creep under the radar and cause excessive heating of the transformer coils. On a US model, this will eventually mean an open primary winding. On Canadian models, which are required to have thermal fuses in the transformer case, it will mean an open fuse (which cannot be fixed except by a partial rewinding of the transformer).

What to do? Scrap those horrid caps on the driver board (and the bias and offset pots while you're in there, and yes...install new electrolytics too, since rising ESR here may well contribute to the problem). What to install? I recommend the most stable 'pf range' caps out there...ones that won't change with age and are as reliable as a brick doorstop: silver mica. On the three that I have rebuilt so far, these are what was replaced -

C1 - 100pf
C2 - 47pf
C10 - 3pf
C12 - 33pf
C13 - 47pf
C14 - 8pf
C15 - 4pf
C18 - 100pf
C19 - 12pf

This is all the pf-range caps on the board (the schematic shows others, but on the board these locations are in fact unpopulated). I was unable to find either a 4pf or 8pf silver mica cap, and bought 3.9pf 1000V ceramic caps. The 4pf was replaced directly with a 3.9pf ceramic, the 8pf has one cap installed on the front of the board, and one on the rear in parallel.

The stock caps were mounted on very long leads, where the caps themselves were about 2" off the board. I honestly don't know the reasoning behind this, but the replacements are certainly NOT mounted that way. I removed the insulating tubing from the leads of the stock caps, and cut short lead insulators for the replacement caps about 3mm long, so the new caps are just a couple of mm above the board, not 2".

I also recommend (as I did earlier) replacing the stock bias and offset trimpots (100 ohm, 2K, and 100K). Simple, single-turn Bourns 3386H-series pots are perfect.

Lastly, deflux the back (solder-side) of the driver board before re-installation. This flux may contribute to the oscillation problems. I use Acetone and a cheap paintbrush. Pour a small amount in the bottom of a cut-off 1 gallon plastic milk jug, and use the paint brush to scrub the flux off. Try not to get any on the front of the board (it isn't a disaster, but leaves a white crust after it dries). If you do, just use a paper towel and wipe it off before it dries. To keep the back of the board pretty, spray with WD-40, smear it around with another cheap paintbrush, and pat dry with a paper towel.

You also need to examine Rf35 and Rf37 on the mainboard. These are 10 ohm, 3W resistors installed in parallel to give 5 ohms, and connected to ground through a .047µf film cap to form the output Zobel filter. If the resistors show any signs of overheating (they're covered with fabric tube weave, but the light-colored weave will darken if the resistors have overheated) the resistors should be replaced, as well as the capacitor.

(A quick mod here while the amp is apart...the output relay is an Omron MY4-02-DC48, a 4PDT, 48V relay. The 48V version of the MY4 is a pain in the ass to find, but you can replace the 48V relay with a 24V relay by changing Rk17 on the protection board from the 820 ohm 1W resistor to a 1.2K 3W resistor. Another quick note...be careful disassembling this amp...it is the Rubik's Cube of amplifiers...and you can easily break the mainboard if you don't know what you are doing.)

Unfortunately, L-07M II #4 might be an iffy proposition to get running again. I have sent the transformer to Gary Brown in Orono, ME. He seems to think he can get it working again, but it isn't going to be a cheap operation. He's confident about his chances, I'm not. I'll be pleasantly surprised if he can get it working. And even if he does, I'll have to put output transistors in the amp that are about 1/10 the speed of the originals.

So...for those of you using a L-07M II that doesn't quite sound right, or are considering buying one, be aware that they have 'issues'. Without proper maintenance it is entirely possible that your amp might be on the fast track to a big, pretty paperweight.

What do I think of the end result after its all done? Damn nice... I'm not tossing my 700M anytime soon, but when everything is doing what it should this amp sounds almost as good as the Sansui BA-F1 or the Accuphase P-300, two of the most transparent solid-state amps that I can recall at this early hour. Well-worth the time and trouble.
 
Thanks. An heroic effort. I was interested because I use the L-05M. It shares the L-0'X'M family's high-speed design parameters. Recently I recapped them (except for the cans) and found no evidence of stress. This model runs VERY cool. It uses a thermatic reference diode in the bias circuit. Perhaps the -07II does too? I found a LOT of flux on the boards that I removed mechanicaly with a dental tool. I LUV these amps, and thus I'd really iike to hear the 07M after your efforts. What about STYROL caps for those pf ranges? I found some NOS here in Sendai. Exactly what were the originals?
 
Well, attached is a few pictures of some of the L07's I have rebuilt. Same experience as EW and you can see the extent I go to with rebuilding these babies. Note the photograph of the interior from the base- wires need to be routed exactly as the picture or oscillation can occur. You can see the original backup set of outputs I have kept for my remaining pair. I have had and restored 9 so far. The L07mkII is an inferior amp compared to the Mk1. PSU caps were dropped from 18,000uF to 12,000uF and the power output actually dropped, although the current limiting is a little less agressive. The bandwidth is specced higher but realistically it is much the same. There are no equivalent devices you can use instead of the outputs if you want the amp to hit anywhere near spec. Power isn't an issue with new devices, just bandwidth. The speaker protectors are fun and games and if yours takes more than a few seconds to trip in- it is likely a leaky transistor and diode on the protector board. Pain to find but solves the problem. The PCB tracks are old and likely overheated, so lifting occurs, particualary in the MkII which used glass epoxy boards instead of phenolic to help the 600kHz bandwidth.
 
pics 1
 

Attachments

  • DSC00718 (Small).jpg
    DSC00718 (Small).jpg
    54.4 KB · Views: 770
  • DSC00207 (Small).JPG
    DSC00207 (Small).JPG
    46 KB · Views: 794
  • IMG_4682 (Small).jpg
    IMG_4682 (Small).jpg
    47.7 KB · Views: 797
  • L07m amp pcb font (Small).JPG
    L07m amp pcb font (Small).JPG
    46.1 KB · Views: 710
  • L07m amp pcb rear (Small).JPG
    L07m amp pcb rear (Small).JPG
    40 KB · Views: 709
pics 2
 

Attachments

  • L07m protector front (Small).JPG
    L07m protector front (Small).JPG
    43.3 KB · Views: 400
  • L07m protector rear (Small).JPG
    L07m protector rear (Small).JPG
    31.9 KB · Views: 407
I love these threads...this one is especially nice, as it validates my memory that the outputs in this amplifier are almost impossible to find...

The only L-07MIIs that I've ever personally located have been smoked in this fashion, and my local tech at the time was unable to bring them back to life...pretty much said I had a power supply connected to an electronics scrapyard...

As always Glenn, an awesome thread.

So would you reccomend any of the above upgrades for the L-07M as well?
 
redcoates7 said:
So would you reccomend any of the above upgrades for the L-07M as well?
Glenn, Is the same problem associated with the L-07M as well? I'm currently looking at one for 'light' duties and am interested if I'm going to have to have it fixed (al la my AU-X1).
 
Did you guys have any luck at all in tracing the source of the oscillation?

I mean some part of the amp's circuit is clearly unstable.

Regards,
Jerry
 
I would be interested in finding out if these maladies are suspect in the L-09's, as well. If so, I may be sending mine to you for the royal treatment. Far too nice to risk letting them go boom.
 
restorer-john said:
The L07mkII is an inferior amp compared to the Mk1. PSU caps were dropped from 18,000uF to 12,000uF and the power output actually dropped, although the current limiting is a little less aggressive.
PS caps in the four L-07M II's sitting right here in front of me are all 18,000µf 71V, and power at clipping on all of them is about 185W. The current limiting in both amps is too aggressive, so neither one behaves particularly well when pushed hard (to be fair, few solid state amps behave well in this region of operation).

I'm unaware of any stability issues with the L-07M, and while the Sanken transistors used in these are very good transistors and difficult to find, they are commonplace compared to the ones in the L-07M II.

I don't know anything about the L-09M. Someone send me a manual and I'll tell ya what I think.
onplane said:
Did you guys have any luck at all in tracing the source of the oscillation?
Like...to a specific cap? I don't see the point. Caps are bad, replace them.
I mean some part of the amp's circuit is clearly unstable.

Regards,
Jerry
Yes...the compensation caps. Ask any amp designer...he'll tell you that an amplifier is essentially an oscillator what you try to 'break' so it'll function as an audio amplifier. Flaky compensation caps can (and do) do this to amps all the time. It's just unusual to see it happen as a matter of course with a particular model. In this case, it is just Kenwood's choice of components that has failed the test of time.
 
Hey Glenn and guys,
Yes, the early L07MII's had 18,000uF and then the later II's had 12,000uF. I have had 4 with original 12,000uF and 5 with 18,000uF's. They also ran the output transistors from the L07 for a while in the MkII's until they changed to slightly faster NEC's. The ones pictured are for the L07m's. They also changed several compensation caps and drilled out tracks on the MkII boards as well as changed the IV protection STV's. There are so many tiny variations in the production run of these it is crazy. All I can say is the best pair of L07's was the first pair in serial numbers for me and I kept them.
The L07m used 2SA747A/2SC1116A (SANKEN) whereas the early MkII's used the same pair, the later ones used (as per the schematic) 2SA1007A and 2SC2337A (NEC) The change in PSU electros was I would say purely cost cutting at the end of their run. If you want a bit more power out of the MkII's- fix the disastrous diode bias clamping part of the circuit as it is pretty much unnecessary, especially if someone has at some stage fitted the wrong STV- mkII's use a STV4H and the MkI's use an STV3H.
 
So why did Kenwood leave their PCBs flux-covered? My 700M had the component side of its PCBs drenched in crusty flux. I thought it odd that a flagship product was allowed to leave the plant with the crap covering the boards.

Maybe I'm spoiled by modern easy-clean fluxes. Back at my last manufacturing job, we could run boards from the wave machine through an old dishwasher and everything came out spotless.
 
I call it flux, but in actuality it is likely a substance to help keep the board somewhat isolated from moisture that could swell and degrade the PC board material.
 
... especially if someone has at some stage fitted the wrong STV- mkII's use a STV4H and the MkI's use an STV3H.
I broke one in one of my L-05M's (despite EW's warning). I found just one NOS STV — a 3H or 4H. I cannot actually remember the original part. Their specs are nearly identical, although the form factor is a bit different. So I am running one channel on one type, and the other channel using the other in the bias circuit. They seem to work fine together. But you might ask, how would I know? Well, it just sounds OK. Installing either one as pair should be OK, especially considering their rarity. Or maybe I am missing something. I've been REALLY careful with these do-dahs since then! I think that I bought the last one on Honshu!
 
The 'STV' components are a 3 diode string in the the case of the 3H and a four diode string in the case of the 4H. They are sometimes called 'varistors' although my definition of a varistor is not a diode string. Doesn't matter, the issue is, some were really reliable, others (the blob variety as I call them- basically three PN junctions covered in a tiny blob of epoxy) were horrible and did everything from go short to open circuit and everything in between at random intervals. Luxman used them in the L80 series of amps and drove me to tears one day trying to find the root of the fault. I don't trust them at all. They are often also used for thermal bias tracking and in the case of the L07's for both bias and overload base current bypassing which they are not good at. I will dig out the schematics for the L07's and post the relevant part.
John
 
L09M uses the same outputs as the L07...
4 2SC1116 (Y) or (O)
4 2SA747 (Y) or (O)

These are preceded by another pair of transistors:

2 2SD381 (M)
2 2SB536 (M)

so there are a total of 4 output devices for the positive half of the waveform, and another 4 output devices for the negative half. Voltage rails
sit at +106V and - 106V respectively referenced to earth ground. I'm assuming that because the outputs are wired in series, that it means each transistor is only dealing with 54 V or so? Is this series/parallel arrangement common for output devices?

Kenwood lists the following substitutions in the SM:

2SD381 (M) - 2SC1913 (Q) or (R)
2SB536 (M) - 2SA913 (Q) or (M)

2SC1116 (Y) or (O) - 2SC1116A, 2SD675, 2SD424
2SA747 (Y) or (O) - 2SB655, 2SA747A, 2SB554
 
FYI, the L-09M also uses one of the "3H" Varistors, STV-3H (Y) in the power amplifier. It's located on the main amplifier board, right by where the daughter board housing offset/bias adjustment plugs in.

A little more digging on the schematic shows it to be downstreams of VRe3... the Bias adjustment pot.

I'm not technical enough to say any more than that!
 
A little more digging on the schematic shows it to be downstreams of VRe3... the Bias adjustment pot.
Downstream? Not UP stream? If it is referencing thermal change, why would it be down? Not arguing ... just wondering.
 
Back
Top Bottom