Simple GK-1 Tweak

PSP · « **on:** 6 May 2006, 07:26 pm »

As the "Pete's building an Orion system" saga continues, I asked Hugh how I might extend the GK-1 bass performance from 15Hz down to 5 Hz (SL recommends that the preamp be capable of 5 Hz, I imagine to minimize phase errors at 20Hz). Hugh suggested increasing C1 to 2.2uf (good film cap) and C2 to 33uf (good electrolytic).

I used a 2.2uf Sonicap (Generation1, $7.60 each from GR Research) at C1 and a 33uf Black Gate NP at C2. I let the system burn in for 150 hours before listening, and base my comments on listening over an additional 100 hours.

This mod clearly improves the bass: there is more definition, more impact and punch in the bass drum, and I am hearing low foundational bass in many CDs that I have never heard before. On a 1 to 10 scale, this would be something like a 2 point improvement.

What really surprised me however, was the improved high end extension and clarity that I heard across the whole spectrum. I listen to a lot of heavy classical (I love Beethoven, Mozart is nice, but a tad too polite sometimes) and--as resolving as the GK-1 is--sometimes the music would get a bit conjested. This mod brought significant improvements in resolution and clarity while simultaneously decreasing the occasional harshness or brightness I hear on some CDs. I was happy with the effect of this mod on the bass, but I am thrilled with the effect this has had on the overall sound of my system.

I used the Sonicap because it has a good reputation among speaker builders and because the cost was reasonable. I have not tried any other cap in this position (other than the stock cap), so I can't comment on what other caps might do. I'm inclined to listen for a while, and then try bypassing my 2.2uf Sonicap with a nice RTX.

Can I suggest that a couple of GK-1 folks here give this a try and report back... just in case what I heard is specific to my system? Since I do science for a living, I like to see new developments replicated by others.

Peter

Jens · « **Reply #1 on:** 6 May 2006, 10:21 pm »

Hi Peter,

I'm just wondering why you didn't use a BG for C1 as well? Any comments?

andyr · « **Reply #2 on:** 7 May 2006, 12:36 am »

Hi Peter,

Yes, I did something similar ... a 10uF BG "N" for C1 and a 22uF BG "NX" for C2.

But I did this at the start, so I can't tell U how it "compared"!

Regards,

Andy

PSP · « **Reply #3 on:** 7 May 2006, 04:29 am »

Hi Jens,
You ask a good question, and I had to think about it for a while... I guess that I'm just biased against electrolytic caps, even very good ones. I believed that a film cap was the more desirable choice, so that's what I used... at some point I should try a BG at C1 and see how it sounds.

When I was building my Foreplay, hanging around the Bottlehead Forum, VoltSecond used to counsel folks to avoid excessively large caps... he felt that they tend to pick up noise by simply virtue of their physical size.

Too many audio projects, so little time...
Peter

AKSA · « **Reply #4 on:** 7 May 2006, 06:27 am »

Peter,

VS (Paul Joppa, by any chance??) is dead right! However, the BG NX 22uF is tiny, less than five mm across, and since it's a can, EM penetration is minimal anyway. The larger RTXs in this category are indeed physically very large, and there is real danger of pickup. In fact, I have experienced this on an AKSA with Andy R when we mounted C1, a very large film cap, too close to the output. Positive feedback caused oscillation!!

Cheers,

Hugh

PSP · « **Reply #5 on:** 7 May 2006, 02:36 pm »

Hi Hugh,
Paul Joppa just posted under his real name. Both Paul and VoltSecond are very, very good.

I will listen to the sonicap for a while, and then put in a BG. I wonder if I can burn in the BG by putting it across the output of an old CDP and let it play for a week. I could solder the cap across a cheap female RCA, plug her in, and let her rip... this should be reasonably close to what the input cap sees? Thoughts?

Peter

Jens · « **Reply #6 on:** 8 May 2006, 06:28 pm »

Hi Peter,

Well, I suppose it will be difficult to compare when I put this mod into my own GK-1R - us being an ocean apart

But please let us know what you think when you swap to the BG at C1

PSP · « **Reply #7 on:** 9 May 2006, 04:33 pm »

To effectively compare a Black Gate NP (say 22uf, since that's what I have) at GK-1 C1 compared to a 2.2uf film cap (currently playing music in my system), I need to find an effective way to burn in the BG cap. There is no way that I can put the BG into my system to break in for 100-200 hours and then trust my sonic memory to make a (probably) very close call.

I want to burn in the caps for a week outside the GK-1 in a way that is reasonably close to the voltage and current the caps will see in the GK-1 circuit. If the circuit is driven by a CD player attached to the circular nodes on the left, the GK-1 input caps see something like this (the dots are just place holders for the diagram, only the dashes and lines represent current paths, the o represents a connection):

o--|C1|---o--Ramp (effective)--o
...............|................................|
.............R1=47k.....................gnd
...............|
o...........gnd

In reality, Ramp(effective) is a frequency-dependent complex number (Resistance + j Reactance), but we probably don't need that much detail here. In a test jig, can I set up the following:

RCA hot--|C BG|
.................|
................Rx
.................|
RCA gnd---o

where Rx is chosen to give approximately the same current and voltage across C BG as C1 would see in the GK-1 circuit. Rx simulates the parallel combination of R1 = 47k and Ramp.

Questions:
- what's wrong with this approach?
- (for Hugh): Can you suggest a reasonable value for R1?

The plan would be to connect the BG cap and R in series across a female RCA and drive it for a week with a cheap portable CD player (2 volts peak). Then I would replace C1 in the GK-1 (now a Sonicap) with the "burned in" Black Gates in one evening and then let the caps break in for 24 hours in the GK-1, finally listening to the system on the second night.

I'm sure that this is not perfect, but it seems like a good way to get close enough so that the methodological errors (in this aproach) are significantly less than the errors due to sonic memory.

Thoughts? Advice?

TIA,
Peter

AKSA · « **Reply #8 on:** 11 May 2006, 11:17 am »

Peter,

Try 15K, driven at line level for a couple of days.

This MIGHT work, but Rabbitz (Peter Jevic) suggested he'd tried this trick and it did not work.

I can't see why it wouldn't work, but then I've not done it this way and can't say with authority.

Cheers,

Hugh

rabbitz · « **Reply #9 on:** 11 May 2006, 01:52 pm »

I didn't run it via line level which I think would have been better as it would be a dynamic load. I only used a sort of cap rack, 3VDC with a diode and a 220R resistor, a static load for over a week. In the end I still had to wait 150 hours.

PSP · « **Reply #10 on:** 11 May 2006, 06:47 pm »

OK.. I'll give it a spin. After the inital listening sessions, I will watch for additional sonic drift in case this wild idea falls short.

I should be able to get this set up in the next night or two, a week of cooking, and then we will see.

Thanks for your thoughts on this!
Peter

PSP · « **Reply #11 on:** 17 Jul 2006, 05:31 pm »

My apologies for the long post, but it's been a long, winding road. I finally have decided that putting a good film cap at GK-1 C1 works well in my system. Because much of this played out while AC was under construction, I will excerpt some e-mails between Hugh and I:

Quote

Hi Hugh, I have replaced GK-1R C1, first with a 2.2uf Sonicap, and then at Jens' suggestion, a 22uf BG NX. I tried to break in the BK outside the GK-1, as described in the above post. That "external breakin" proceeded for almost 3 weeks, 2 weeks attached to the tape out loop of an old cheap reciever and one week attached to the analog out of a Sony DVD/CDP.

On Monday night, I placed the "broken in" BG NX cap at C1, replacing the Sonicap. The sound with the BG cap was thin and bright. After 48 hours of burn in, there is no major change in the BG. I have the BG oriented with the long lead toward the input pin... is this OK?, and is there any likelihood that BG orientation could cause a significant change in sonics?

My intention right now is to burn in up to 100 hours and evaluate, unless you suggest swaping the BG lead orientation.

Many thanks
Peter

Quote

Hi Peter,
The large foil polyprops have such voluminous geometry that in the confined pcb of the GK1 you could experience hum and RFI intrusion. The very small 22uF NX is probably as good a cap, but because of its small size, there's much less likelihood of these problems, AND the higher capacitance lessens phase shift at very low frequencies, which is alleged to have sonic benefits (though I've not tested this).

The longer lead should, if one assumes this is half way positive, go to the preamp side; the shorter to the signal input side.

You cannot expect much change in the first couple of days. I'd say swap around the cap, and break it in for at least ten days. Then have a critical listen.

To summarize my results at this point, I found that the 22uf BG NX at C1 sounded pretty terrible, but--as Hugh points out--I probably had the cap leads oriented backwards. The NX is nonpolar, but one orientation apparently sounds better than the other. My external break-in experiment may have worked (I didn't hear any continuing changes in the days after putting in the "broken in" cap), but since I had the orientation wrong and it sounded bad, it might be really hard to tell if there were sonic changes.

Following Hugh's advice, I put in a new 22uf BG NX at C1, this time with the shorter lead toward the input pin. I used a new cap (rather than the "broken in" cap used above because I wanted to be dead certain regarding lead orientation. I let this new cap break in for 12 days of essentially-continuous playing (mostly classical music from a local FM station) with interspersed listening in the evening. Before starting this experiment, I had listened to a few key selections and refrained from listening to these until I was ready to do a formal evaluation. We went away for the weekend (over July 4th) and I returned to find that one channel of my GK-1 had been disabled, apparently by lightning (see Bolt From the Blue post for more info). Working with Hugh, we replaced T3 and T4 in the bad channel and got music playing again. After getting it fixed, and an additional 100 hours of burn-in I spent two evenings doing some serious listening. My report to Hugh and Jens:

Quote

Gentlemen,
I listened hard to the 22uf BG NX at C1 and it was hard to find fault. After 12-14 days of accumulated burn-in time, with a lightning strike mixed in for good measure, I thought it was perhaps still a bit rough on complex passages with massed strings in the upper registers, high brass, etc., but really it was pretty good. Definitely nothing to complain about.

On Saturday I had a few extra hours, and decided to put the 2.2uf Sonicap back in, just to do a fair comparison before drawing conclusions. It took about 15 minutes playing cuts from Jens' My Music to be pretty sure and another several hours last night listening to some of my best CDs (for example, http://www.referencerecordings.com/minnesota.asp#rr95 ) to be sure. As I told my wife, it was hard to complain about the 22uf BG, but I am in love with the Sonicap.... to my ears, it is almost magical in its effect. The instruments are more real, more vital, colors are brighter, textural contrasts more palpable. Even though it is more resolving in this position, the Sonicap is clearly smoother too.

I did not notice any difference in noise (just paused the CDP and listened to the speaker cones) between the two caps. I mounted the Sonicap (diam ~ 16.5mm) vertically on the component side of the PCB... I had to kind of wedge it in, but I didn't have to move any components to fit it in.

So, based on my experience, I think it might be worthwhile to try a good film cap at C1. I'm not advocating for Sonicap... it was simply the first film cap I chose (because speaker builders seem to think it's very good and they are reasonably priced at $7.60 each). I would be eager to hear what you and others might find, but I will probably stick with the Sonicap unless a consensus emerges that something else is a lot better. This cap at C1 was such a big step forward that I'm thinking I might bypass them with 0.022uf Relcap TFTs to squeeze out the last drop of performance here.

I will post these observations on AC in the next few days.

Now I will move on to building the Orion...

Take care,
Peter

After listening to the Sonicap at C1 for another week or so after writing the above note (the Teflon caps have not arrived yet), I am still extremely happy with the sound, re-discovering most of my CD collection again.

Peter

SamL · « **Reply #12 on:** 17 Jul 2006, 09:54 pm »

Hi Peter,

Not sure if you come across Tony Gee caps test = http://www.humblehomemadehifi.com/Cap.html
Intresting read and you might want to try bypassing with Vishay Roederstein MKP1837.

Have fun,
Sam

PSP · « **Reply #13 on:** 19 Jul 2006, 10:50 pm »

Hi Sam,
I had seen that web page, but perhaps a couple of years back when there were a lot fewer caps reviewed. In any case, I had forgotten about it. I've printed it out for my paper files and put the page on my Favorites.

Thanks!
Peter

PT914 · « **Reply #14 on:** 20 Jul 2006, 07:56 pm »

Hi Peter,

I just pulled the 22 uf BG NX cap from C1 and replaced it with a 4.7 uf Solen cap. You're right, a film cap here sounds better.

Thanks,
Philip

AKSA · « **Reply #15 on:** 20 Jul 2006, 11:36 pm »

Y'know,

I think Peter and Phil are right; the NX is very good, but I believe a good quality film cap sounds better, somehow more 'natural'. (now there's a scientific term!)

The LF has gone back to a film cap in this position. Now the search is on for a small, low voltage filmcap which sounds exceptional!

Cheers,

Hugh

jules · « **Reply #16 on:** 23 Jul 2006, 12:22 am »

Peter, Phil and Hugh,

this C1 cap does keep popping up doesn't it. I think Hugh first suggested that a Rel cap in this position was very good several years ago [I remember this as being even before the adoption of BGs although I might be wrong].

There seem to be a number of options here but the price goes up exponentially according to size. It seems that the 2.2uf is ok instead of the 22uf BG though [in terms of capacitance]. Still, the question is, what is it worth trying? A Relcap exotica comes in at a cool $250

. An RTX is about $40 with Hovlands, TRT dynamicaps and Jupiter beesewax slightly less. What version/size of Teflon are you going to try Peter? Does anyone have any opinions about these choices, based on experience in other situations that could be useful?

While we are talking GK-1 in this topic, it sounds as though the BG at C1 in the 55N+ might also benefit from a change [and Hugh has raised the question for the LF].

Since microphonics don't seem to be a problem is size important other than for physical reasons?

jules

AKSA · « **Reply #17 on:** 23 Jul 2006, 12:33 am »

Jules,

On the AKSA, a minimum of 1.5uF is required, but on the LF 470nF is fine, since the two circuits have very different input impedances.

On the AKSA/LF, the very large foil caps, notably the RTXs, induce positive feedback if mounted under the pcb, adjacent to the output inductor. This causes destructive oscillation, and must be avoided at all costs.

Since there is little space on the component side of the board, the best option is to short the two cap pads on the amp module and run a shielded wire back to the RCA at the rear panel of the amplifier enclosure, and connect the 'hot' signal input via the large cap, which then connects directly to the RCA center post. This keeps RFI to a minimum with a shielded cable, and moves the large cap to a spot some distance from the speaker output wires/binding posts, leaving the module looking pristine and neat.

I used the BG 22uF because they are very small, sound very good, and their high capacitance improves measurable phase shift at very low frequencies. The AKSA benefits from this; the LF not so much as its bass is stronger. And of course, keeping all these NX caps the same size keeps inventory under control.

A 470nF RTX is much smaller, of course, but still too large to mount on the pcb. Someone might try the X7R low loss ceramics; they are reputedly good sounding, Ginger has reported this.

Cheers,

Hugh

jules · « **Reply #18 on:** 23 Jul 2006, 01:12 am »

thanks Hugh ... brilliant!

I might read this back to you at some point just to make sure I've got the implementation right before I try it.

Jules

PSP · « **Reply #19 on:** 23 Jul 2006, 08:02 pm »

Jules,
I bypassed the 2.2uf Sonicap I used for C1 with a Relcap 0.022uf 200v TFT (dimensions 0.23" D x 0.75" L), roughly $USD 20 each at Parts Connexion. I chose that one because I wanted to mount the bypass on the foil side, under the C1 Sonicap. The TFT was small enough to fit between the GK-1 Analog PCB and the top plate of my chassis.

This cap is still breaking in, but (after 100+ hours) it is sounding like it might be a keeper. My intention is to listen to music now... if the AKSAphile community finds another cap for C1 that is just wonderful, I may give it a try someday, but--in my system--a 2.2uf Sonciap ($7.60 each), possibly bypassed with a really nice smaller cap ($$$), sounds very, very nice to me.

Take care,
Peter