Possible improvements to Luxkit KMQ7 circuit

Keego

New Member
Hi All,
I joined the forum recently and have found it to be a gold mine of information in so many different departments. I'm hoping to get some help improving the performance of my Lux power amp. it is a push-pull amp with KT66's as outputs and a 6an8 driver section. I believe originally the design had 6L6's as outputs. I have a reasonable understanding of how my amp works, and have done a lot of work on it over the years, re-capping, new output transformers (one original failed), new improved individual bias circuit, balancing pots for 6an8's etc., and I am very happy with the sound, apart from some harshness in the upper mids, which I am fairly sure is being caused by the amp, as it is present on phono and cd sources (passive preamp to power amp).
I managed to find a circuit diagram online, and have uploaded it here if anyone wants to have a look. The NFB circuit seems a little unusual as there are electrolytics in the signal path, which I believe is not the greatest idea from a sound point of view. Also, there are no stoppers on the output tube grids, and I recently put some carbon comp Allen Bradley 1.5k's I had kicking about on the grids, and to me the top end sounded noticeably smoother as a result.
Just looking for advice on any improvement I could make without radically altering the circuit, because as I said I'm very happy with the overall sound, just looking to get it as good as it can be. Here is a link to the circuit diagram online if the jpeg is unreadable: http://music.geocities.jp/hoppyfather/page131.html

Thanks,

Keego
 

Attachments

  • img616.jpg
    img616.jpg
    49.6 KB · Views: 66
Last edited:
The electrolytic caps at the 6AN8 are cathode and screen bypass caps. They aren't really in the audio path.

Circuit looks very much like the Dynaco ST70 or Mark III. Not quite identical but the same basic driver design and a pentode output stage.

Does it sound better with 6L6GC's instead of the KT66? Off the top of my head I want to say the KT66 isn't rated for quite as much abuse as a 6L6GC but I don't have a datasheet in front of me to confirm that.

Not sure if the beam plates are internally connected or not, but it would be worth confirming. If they are not, you'd need to add a jumper to make that work.
 
It had KT66's when I bought it a long time ago, so I've always replaced with same, currently a Gold Lion matched quad. I think the 6L6 is rated at 30W and the KT66 at 25W. Doesn't really matter for me as my Ls3/5a speakers can only handle 25W max anyway, and I never push them that hard.

Is it unusual that there is not a small value capacitor in parallel with the 4k resistor in the NFB circuit, I always though there should be one there to tame ultrasonic frequencies that could cause instability, or am I misunderstanding the purpose of that capacitor?

Thanks for your help!
 
That cap that is typically parallel to the feedback resistor is quite dependent on the output transformer make and model used. It may be that the designer found that one wasn't necessary, or it added little value. Also note the cap and resistor series combo hanging off of the grid of the phase inverter to ground. That is there for taming ultrasonic frequency response for purposes of making a more stable amp. It may be that the values used there offset the need for a phase cap across the feedback resistor. But plug in a different make output transformer, even one with the same primary and secondary impedances, and those values would need to change, and a phase cap across the feedback resistor might then be necessary.
 
Just looking at schematic on my phone ( not super sharp though)
Thain, you weren't kidding. Looks like a Dynaco clone
 
That's interesting. I did change output transformers many years ago, but have no idea how they compare spec. wise with the originals.. They are made by Sowter in the UK, and I know they are excellent quality, but the amp has always sounded a little "toppy" to my ears; not a lot, but a little. I wonder if experimenting with some different values across that feedback resistor would be worthwhile?

Of course, the best thing would be to put a scope across it, but I don't have access to one. Another point, I wonder if that 4k resistor is the correct value for the current opt's? Probably not...!
 
If you didn't change the secondary tap point from which the feedback is connected, then the 4K will be correct--if you wanted about the same amount of feedback as the original design used. On the other hand if you changed the feedback tap point on the secondary (say from the 16 ohm tap to the 8 ohm tap), then the feedback resistor would need to also change, if you wanted to keep the feedback amount the same as in the original design.

Exactly how much smaller or larger that feedback resistor needs to be to compensate if you change feedback tap points depends on which tap you moved it to relative to the original design, and the size of the lower resistor in the cathode circuit of the first stage where the feedback connects. If that lower cathode resistor is less than say 10% of the size of the feedback resistor, then it can be pretty much ignored in the feedback resistor size calculation.
 
The scale factor of the feedback resistor is roughly (if lower cathode resistor can be ignored): square root of (new impedance tap divided by original impedance tap). Thus if moving from 16 ohm tap to 8 ohm tap, feedback resistor scale would be sqrt(8/16) = 0.707. So the feedback resistor would need to be 4K * .707 = 2.8K, and 2.7K would probably be close enough. This is an approximation depending on the size of the lower resistor in the cathode circuit where the feedback connects.
 
I strongly suggest you add screen stoppers. 100R should do the job.

I think most of us would argue that the 100uF cathode bypass caps are in the audio signal path. Since they're effective down to about 1Hz nominally, I would try replacing or at least bridging them with 1000uF or more. Douglas Self has shown that 'lytics need to be big enough to keep signal voltage across them below 80mV for zero distortion.

Replacement OPTs make it almost certain that some high-frequency phase/frequency tweaks are needed, but a squarewave generator and scope are essential for this work. You're actually pretty lucky to have avoided massive oscillation.
 
well, "not in the audio path" meaning that the amplified signal doesn't directly pass through them. They aren't a coupling cap. They are important though.

Unfortunately debates about precisely what is and what is not audio path can get fairly involved.

What was the reason for the output replacement, were the originals bad?
 
It's a long time ago, but if I remember correctly, there was a massive short in one of the output tubes, a really bad smell and afterwards the primary side of the opt concerned was open circuit..

I think I need to get someone with a generator and scope to have a look, I have someone in mind. I will try and get my hands on some 1000uf caps as well. Thanks for all the help guys, much appreciated.
 
Thats unfortunate. I've heard one pair of Luxman mono amps, not this model, but they were very nice sounding pieces of gear.
 
I strongly suggest you add screen stoppers. 100R should do the job.

I think most of us would argue that the 100uF cathode bypass caps are in the audio signal path. Since they're effective down to about 1Hz nominally, I would try replacing or at least bridging them with 1000uF or more. Douglas Self has shown that 'lytics need to be big enough to keep signal voltage across them below 80mV for zero distortion.

Replacement OPTs make it almost certain that some high-frequency phase/frequency tweaks are needed, but a squarewave generator and scope are essential for this work. You're actually pretty lucky to have avoided massive oscillation.

I was just wondering, what is the purpose of that 100mF cap in this circuit? Looking at schematics for Dynaco mk3 and other similar designs, the feedback loop seems to terminate directly to the cathode bias resistor. Could I get rid of this capacitor, and is it having a negative impact on the sound of the amp? My knowledge is a bit limited on this matter I'm afraid..
 
The 100uF cap is a cathode bypass for the voltage amplifier stage. It shunts AC around the 1.5K cathode bias resistor and thereby prevents loss of gain due to cathode degeneration. The 100R resistor below it is the low side of the feedback voltage divider. It is also in series to the cathode and does have a small effect on gain. Removing the 100uF cap would dramatically reduce loop gain and feedback, but shouldn't do any harm, in case you like to experiment with stuff like this.

"100mF" means 0.1F or 100,000uF in modern times.
 
Last edited:
Haven't dug in too deep, but ... is the addition of CCS (constant-current-source) an option? That's a killer mod popular with the Dynaco crowd, providing major improvements in overall sound quality ...

Doesn't seem that it'd be all that difficult to implement either ... basically, you're just replacing a fixed resistance value with a "smart" circuit for better signal level control? My VTA driver board adds two LM334's and a couple extra resistors to tweak the idle current to do the job.
 
Haven't dug in too deep, but ... is the addition of CCS (constant-current-source) an option? That's a killer mod popular with the Dynaco crowd, providing major improvements in overall sound quality ...

Doesn't seem that it'd be all that difficult to implement either ... basically, you're just replacing a fixed resistance value with a "smart" circuit for better signal level control? My VTA driver board adds two LM334's and a couple extra resistors to tweak the idle current to do the job.

I actually fitted a VTA driver board to this amplifier, took a bit of thinking about because its obviously not a drop-in replacement like on the ST 70. However, I just didn't like the sound of it, and ended up re-installing the 6an8 driver stage.I experimented with a range of NFB settings but just found the sound sterile and uninvolving. Very subjective I know, but it just seemed to rob the amp of all its magic. Made some small improvements to the old circuit and it sounds better than it ever did now. Might keep the VTA board for future project though..
 
Seems to me there's a new octal version of the VTA board available now as well.

I ended up using a 6SN7 in the center position here, using an off the shelf socket adapter, making it a simple plug 'n pray. Iron's more than strong enough to handle the extra demand. I'd think three of those might get a bit crowded, but I understand some have tried it an had good results. I went with 5963's in the side drivers, for a very complimentary trio.

I also had to add a "socket saver" riser to clear the hole in the top of the custom top plate. Another couple millimeters, and it wouldn't have been a problem. Anyway, ended up with that whole "bowling trophy" look, eh. I can live with that.

6sn7-adapted.jpg


PS ... if your iron and layout will handle the TungSol KT120's, I give those a big thumbs up as well for a simple tube roll. Outstanding tonality, especially at the bottom, and they're actually cheaper than a couple KT88 options that are now relegated to backup duty.
 
I just picked up a KMQ-7 last week as well; I realize this maybe sort of a way too simple question.. but where are the bias measurement points on this thing? Mine has the 6L6GCs in it.

Any "preventative maintenance" recommended? Seems like the coupling caps have been replaced at some point.

I do notice this amp is a little harsh up at top at times.. at least in comparison to my MB3045s. The bass is rather muddy too.. might be due for new set of power tubes to start.
 

Attachments

  • IMG_0298.JPG
    IMG_0298.JPG
    117 KB · Views: 22
As built, this model has no bias test points and requires a matched quad of output tubes. I imagine that bias was tweaked at the factory for a specified power line current measurement. At minimum, it's best to add cathode current sampling resistors of 1.00 or 10.0 ohms, 1% tolerance, so you can get a handle on each tube's operating point and do your own matching.

If you're opposed to modifications of any sort, you could check bias by measuring OPT primary DC resistances and then checking voltage drop across those resistances with the amp in operation. But don't try that unless you're well acquainted with high-voltage safety practice.
 
Back
Top Bottom