Electrolytic to film - Retain the bypass in coupling app?

Hi everyone,

I have a question that I know someone on the late night crew can probably help me with.

Say I have a certain commercial preamp that has a 50V 22uF bi-polar electrolytic bypassed with a .01uF poly film and I want to remove electrolytics in the signal path.
I have (5) 4.7uF WIMA M.polyester that I can parallel for about 23.5uf of film cap to get rid of the electrolytic.

My question is, do I need to retain the .01uF poly film.
I assume the polyesters are good to the higher frequencies that the electros are not - and bypassing is normally avoided when it is not necessary...

Any help or suggestions are greatly appreciated :)

Hafler
 
HaflerFreak said:
Hi everyone,

I have a question that I know someone on the late night crew can probably help me with.

Say I have a certain commercial preamp that has a 50V 22uF bi-polar electrolytic bypassed with a .01uF poly film and I want to remove electrolytics in the signal path.
I have (5) 4.7uF WIMA M.polyester that I can parallel for about 23.5uf of film cap to get rid of the electrolytic.

My question is, do I need to retain the .01uF poly film.
I assume the polyesters are good to the higher frequencies that the electros are not - and bypassing is normally avoided when it is not necessary...

Any help or suggestions are greatly appreciated :)

Hafler

Just a guess, but you probably wouldn't want the frequencies that the .01uF would pass, anyway. I could see 0.22uF or even 0.1uF, maybe, but not 0.01uF. There should be a lowpass RF filter, for the input, anyway, which would probably remove whatever a 0.01uF would pass. I'd at least try it with no smaller bypass cap connected.
 
Hi Tom. Always nice to meet other nocturnals ;)

The input cap is a SEACOR 250V wound metallized .47uF polypropylene. It is stock, the only thing I have that I could swap is a Panasonic ECQ-P or a 1uF Radio Shack film that is highly recommended. I doubt I need to go to the trouble, the Seacor has always sounded fine and it is stock - that's usually good with a Hafler.

I suspected that the .01 was unnecessary with the change from electro to polyester. Film at any value works to higher frequencies than any electro.

Thanks!

I will post immediately if I hear any significant difference over the electrolytics. Cross your fingers...
 
HaflerFreak said:
Hi Tom. Always nice to meet other nocturnals ;)

The input cap is a SEACOR 250V wound metallized .47uF polypropylene. It is stock, the only thing I have that I could swap is a Panasonic ECQ-P or a 1uF Radio Shack film that is highly recommended. I doubt I need to go to the trouble, the Seacor has always sounded fine and it is stock - that's usually good with a Hafler.

I suspected that the .01 was unnecessary with the change from electro to polyester. Film at any value works to higher frequencies than any electro.

Thanks!

I will post immediately if I hear any significant difference over the electrolytics. Cross your fingers...

<grin!> Hi. Yes, I'm half-nocturnal, lately at least. It's probably an hour later, here, though (~3:30 now). So it's now officially past my bedtime. :)

Fingers crossed.
 
Hey AndrewT,
In fact, I have read and reread that thread with great interest. Thank you for adding the link to this thread (I found it researching EPCOS caps...).

The thing is, they are talking about input coupling caps.
The stock input cap on my unit is a 250V .47uF MPP that I do not have immediate plans to experiment with, but if I do, that thread has me very curious about the Radio Shack 1uF films.

What I am hoping to do is replace an output coupling cap arrangement that consists of a 22uF 50V BiPolar electrolytic bypassed with .01uF 100V MPP.

In case anyone would like to know, it is a DH110 preamp and I am talking about C21 and C22. Schematic at Hafler.com

Except for a couple huge 10uF MPP caps, the best alternative that I have found are little WIMA MKS2 50V 4.7uF with 5mm lead spacing. I can parallel 5 of these easy in the space that I have.
I should be doing this some time in the next 8 hours or so, I will post what I hear.
 
Hi HF,

IMO, Input and Output caps basically do the same thing! They block unwanteds yet allow the wanteds in a balanced, pleasing sort of way at the handoff to/from another component!

Forget the 5mm spacing... that precludes 99% of your possibilities. We don't do this for convenience but rather to improve our listening enjoyment. You should see what AndrewT suggested I do with my CD Player's analog output stage. It doesn't look anything like the PCB and it is a great way to handle my oversized caps and some other improvements.

Also, look at some of Extreme_Boky's work he has posted with pictures. That will show the level of creativity some of these guys have to go through to get the most rewarding results.

Research what qualities different caps bring forward (there is a lot of info about this here) and narrow down to a few choices for what you are looking for. Then set up a controlled trial/listening/comparison test (don't forget to burn in the caps first or you won't be able to compare fairly). This is how you will find the best choice or choices if you choose to bypass. Unfortunayely there is no way around it. This is a very personal matter and things have changed quite a bit since that Hafler was manufactured. Now, there are more modding choices than you can imagine!

I hope you will actually enjoy it! Nothing is more satisfying than listening to your system bloom right before your ears as you make great mods and choices! But to get there, you have to do your Due Diligence!

Also, if you find a place to get Dweekie's Epcos 4.7uf caps, please let me know. I want in, I need four!

I have a DH500 that I used from 1984 until last Spring! Maybe I will mod it someday. I hear it has potential!

BTW, my kitten/cat is Nocturnal, would you guys mind if she posts on your threads late at night? It just may keep her out of trouble! Well at home anyway, who knows what she will stir up around here! LOL

Good luck! Keep us informed!

Regards//Keith
 
Thank you for your generous reply, Keith, great to hear from another Haffie owner. Very cool of you to recommend some of the other posters' projects. I will be checking into those after this post.

Nocturnal cat... :blackcat: :sleep: Hilarious!

I have not performed my mod yet, but I was just about to sit down and solder my MKS2 caps together so I thought I had better check this thread.

From what I have already read on the board, I know just about anything beats an electrolytic in the signal path so polyester or propylene is not such a big deal. I know polypropylene is ideal, but I figured I would give the MKS2's a try because: 1. They're film and 2. They're small (5mm like I had mentioned.)

If I parallel 5 of them and stack them on their sides, vertically, they will sit right on the stock circuit board using the space and original holes for the electrolytic.

One thing I have been wondering about is whether I could go with a smaller uF value when switching the electro to film. To tell the truth, I am looking to pick up a little in the bass floor, so I thought the 5 4.7uF caps at about 23.5uF would be a good replacement for the 22uF electro.
My two 5 cap groups have been selected with a precision cap meter to have matched sums of 24.22 and 24.19 uF - closest I could get.
 

Attachments

  • wima47.jpg
    wima47.jpg
    62.9 KB · Views: 570
When you finish installing these, leave the Preamp on with a FM tuner hooked up to it with a strong station for the weekend (no amp required).Those caps will take at least 50 hours to get closer to sounding what they will settle to. No judgements before will be valid and they may continue to change up to 250 hours of use!

Not just about anything beats a Electrolytic. The Blackgate N and even better NX series have some following for a reason. They are worthy competition and some times can out perform a film cap, especially when bypassed properly. But the only way to find out is to try different caps and combos (bypassing). See what Dave is doing in the Tripath thread!

If you post the schematic of the output stage of the Hafler, the guys might be able to suggest different values (lower hopefully)for the Caps and (maybe higher) resistors (metal film) and tune up that output stage. Things are different since the early 80s. Also knowing what you are hooking it up to (Amp) might help to tune it for your situation!

Luna will be logging on sometime in the middle of the night. She always gets a burst of energy after midnight that lasts for several hours. Then she is up before us! How does she do it? Must be the cat food.... I'll have to try some!

Regards//Keith
 
RATS! Oh life, what am I going to do with you....

As usual this project has run late (1 AM by me) because of unscheduled interruptions and humbug.

I have finished my 'supercaps' and am attaching a pic. For anyone that noticed, I did match the caps closer by flipping a couple -- 24.2 and 24.21.

Once finished they tested 23.3 each on a lower resolution scale.

I will attach the schematic soon enough as I would love advice concerning the muting JFETs. There are also (2) 4.7uF bipolar electros in that circuit that could use these WIMAs.

Concerning Blackgates:
Where the heck do you guys get these things? I have looked everywhere and cannot find a source for Blackgate caps - or many other high end caps.
Digi-key carries ELNA SILMICs but those are polarized electros for PS use.

To be honest, this is already a bit of an ongoing experiment. The original 22uF electros were IC caps. They sounded great and I still have them if I need to solder them back in one day.
The caps I am listening to right now are XICON non-polar electros and, to be honest, they sound quite good except that I would like a little more presence in the bass and mid bass. The XICONs have surprised me, though.

I read somewhere that electros can pass bass better than films - is this true? Anyone have an opinion?

Say 'Hey' to Luna, I really miss all of the cats and dogs I have had or lived with through the years. The best cat I ever knew was a runt born with one eye. She was awesome beyond words - her name was 'Kitty'.
 

Attachments

  • supercaps.jpg
    supercaps.jpg
    57.4 KB · Views: 564
Here is the DH110 schematic link:
DH110 schematic

To Keith:
My amp is direct coupled. No input cap whatsoever. It is a Hafler XL280, and is in fine working order. Am I gathering correctly that the output caps on the preamp are mainly to block DC from backing up into the preamp and thus largely unnecessary in this situation?

I don't plan on eliminating the output caps, just finessing them so to speak.

I don't know if anyone will be able to make it out, but I will attach a lowres pic of the caps in the circuit...
 
Hi HF,

See, creativity.... I like how you stacked the Wimas and soldered them to two "poles" and taped them to make them stable. This is one of those places where solid silver wire is a great idea. I hope you used heat sinks on the legs of the caps (very tempermental with heat).

Input/Output caps can block DC both ways and are redundant in most equipment connections. Finding transparant ones that don't influence too much twice, is hard (output and then input), so if you only have them at one end it can be better but fine tuning takes time. But that makes swapping equipment a project.

The schematic picture is clear. Hopefully some of the guys can make suggestions to some of the values of the different caps and resistors for best sonic results.

When you ask about the Bass and Electrolytics vs Films. Remember that when you get to a certain cap or other component in a circuit, it would be nice to think the signal is perfect to that point. However it probably isn't. Every component (most) has its own signature, however subtle it may be (sometimes negligable) and some more than others, IMO. So when the designers do these circuits, they keep compensating the signal to bring it back to the middle (keeping cost and availability in mind), where it is supposed to be. So with every circuit different, the only way to know if you are getting the effects you desire is to try and listen. Each component may have a reputation to present a certain way but this is in general. Some circuits don't let it present the normal way. So my boy... Back to trial and error but with a wiser sort of way.

DaveT ("Tripath Input Coupling Caps" thread) is using Michael Percy for Blackgate N and/or NX series caps at 4.7uf. They have a $25 minimum. Also on his thread is a "Burn In Circuit" that you can make for under $10. It will allow you to burn in your caps (many at a time if necessary) so when you listen you can make fair comparisons because most caps change with more hours up to a certain point. Each cap has its own time required. Some actually start out sounding obnoxious and then settle to something wonderful. If you install caps and wait until they settle before trying new ones and waiting until they settle, you won't get the impact of the change. MAKE ONE! You can get what you need at Radio Shack!

Maybe one of the Moderators can help you change the name of this thread a little so you can solicit Hafler Preamp help! This thread needs a lot of traffic for you to get some brilliant ideas! Hafler experience might be additionally helpful.

Luna says Meeee Owwww!

Regards//Keith
 
Hi again, HaflerFreak!

My cat is named Scaredy. :) She's jet black. I've had her since 1992 or so. She likes to stay outside (I'm outside of town, with a little acreage, woods, etc. It's fine.). She's a great cat. [My dog, on the other hand, likes to stay indoors. And she's afraid of the cat.]

I 'm very interested in hearing about the sound of your system, with your new capacitor assemblies, especially after they've been burned-in.

-----

It might be interesting to know what you have for C23 and C24, the power supply decoupling caps, and to see the actual circuit layout.

You probably know all of this, already, but...

One way I like to think about it is that the original "signal" doesn't actually make it past the first active device. Each transistor stage pulls its output signal from the power rails, not from its input. The input only controls it. So, at every stage, the "signal" going to the next stage is really coming from the power supply rails. There is no doubt, then, that the power supply is "in the signal path". And you can think of the decoupling caps as the power supply's 'local representatives'.

The power supply decoupling capacitors might 'normally' be thought of as only for filtering 'glitches' and noise out of the power supply rails. But they should also at least be thought of as 'local quick-supply reservoirs', to better-supply current on demand, for each transistor or transistor stage to use to create its output signal.

A local reservoir is needed because the wires or pcb traces from the power supply have parasitic inductance and resistance, which would otherwise impose their meddlesome, muddying transient response on whatever dynamic currents the transistors tried to draw, and might also cause other problems.

At any rate, because the actual power supply is usually separated from the current-using devices by the conductors' parasitics, the main part of the power supply that should be considered to be 'in the signal path' is the 'local' part, i.e. the decoupling caps and the conductors from them to the devices that need to draw the current from them.

If we were expecting only one frequency and amplitude at the amplifier input, it might be possible to 'tune' the decoupling caps' values, so we'd always get exactly the right currents at the right speeds, from them (which might also mean that the main power rail, upstream, would see no disturbance whatsoever, when we drew those currents).

But, with music signals, we just have to try to make sure that the decoupling caps can supply enough current, at the right frequencies/speeds, for whatever peak demands we might expect. If not, the new signal formed from the power supply current, at the output of each stage or transistor, might not be faithfully reproduced, according the the stage's control input.

Besides the characteristics and the actual capacitance values of the capacitor or capacitors used for power supply decoupling, another thing is often at least as important: As they say in the real-estate business, "the three most-important things are, location, location, and location". It basically comes down to the parasitic inductance and resistance of the wiring or pcb traces, again. Even a theoretically-perfect set of decoupling caps can be well-sabotaged if there is too much conductor-length between them and the device(s) they feed. For that reason, each device that draws significant or fast-changing currents should ideally have its own decoupling capacitor(s), connected as close as possible to its power supply input. [There are also possible problems due to upstream conductor parasitics and components. But they are too invlolved to get into, here. See thread link, farther below.]

I've gone on for far too long, already. I was just curious about whether you'd yet considered modifying the decoupling capacitors' configurations. For example, larger caps might supply lower-frequency current demands better, but might not be as good for higher frequencies. Adding some smaller/better caps in parallel might help re-balance the frequency response, but might also tend to introduce high-frequency resonances (depending on those pesky parasitic inductances and resistances), and might need to be snubberized with a small series resistance.

Technically, I guess what you'd want would be for the impedance of the power supply rails, as seen by the points of load, to be as small as possible over the frequency range of interest, and to not have any undesirable 'resonance' peaks.

There are technical/mathematical ways to figure it all out. But it's usually somewhat difficult to get the data needed, about all of the parasitic inductances and resistances (although they can often be estimated, 'fairly easily').

There is a very good thread about paralleling electrolytic and film caps, at:

http://www.diyaudio.com/forums/showthread.php?s=&threadid=106648

There are two different simple formulas that posters derived, there, for the optimal film cap value to use in parallel with an electrolytic, depending on the caps' values, ESRs (equivalent series resistances), their ESLs (inductances), and the parasitic R and L from upstream.

In post #146 of that thread, I gave links to some good papers about snubber design, at least one or two of which might be of practical benefit. But, sometimes, all you need to do is add a small resistance (or ferrite bead), just upstream, in series with the power supply rail, and you can then probably parallel almost whatever you want.

I'm sorry to have blathered-on, for so long, about all of that!

I guess the bottom line is that you get to experiment with the decoupling caps. :))
 
Excellent info, thank you Keith and Tom.
To Tom, trust me, you weren't blathering for two reasons:
1. I have been nicknamed 'motormouth' in the past.
2. What you had to say was intelligent - you cannot imagine the inane, ignorant blather that I must endure at my job. :crazy:

Unfortunately, I work long hours over the weekend so the WIMAs are on hold for a day or two. I am actually using my time to study the sound of my current system with its XICON nprl caps.

Indeed I have experimented with my decoupling caps. The results have led to multiple changes because each time, I lost qualities in the music that had been there with the Hafler original IC caps. My bass issue could definitely be here somewhere.

I found mostly frustration but here are the results so far:
Local Decoupling caps are (8x) 25V 470uF
Upstream main PS caps are (2x) 50V 1000uF
bypassed with (2x) 50V 22uF and .1uF 100V polycarbs


Local decoupling:
1. ELNA SILMIC - Strongest bass of any cap that I tried but it sounded, quite literally, like there was a pillow over my tweeters.

2. Panasonic FC - Clear, a little more 3D than the ELNAS, but no bass and still a problem with restraint in the highest highs.

3. XICON Low ESR (ESRL series) - I just know the mention of XICONs must make some cringe on this board, but these are the only caps I found that were even in the ballpark with the original IC caps.
These are the caps I am currently listening to.

Upstream PS:
Once I started making changes, nearly everything I tried was detrimental. In the schematic, Hafler calls for Low ESR local decoupling caps, but the main PS caps are general purpose. I decided to try switching the 1000uF and the 22uF caps to low ESR Panasonic FM types. Yuck! No spatial qualities, no bass, no life.
Then switched to Nichicon HE - same thing.
The original IC caps were gen. purpose 1000s and high temps (105) for the 22uF. I decided to mimic the performance specs of the original parts in hopes of original performance, so I went with Nichicon VR/VZ in stock values for the main PS.

This has been acceptable and are the caps I am currently listening to.

Now I have blathered on, but I hope this info was helpful for other experimenters...

I will try to update in the next day or two.

Thanks again to Keith and others for some great info/posts.
 

Attachments

  • kitty.jpg
    kitty.jpg
    69.9 KB · Views: 332
Listening tests: Electrolytic vs Polyester

Man!!!!!! that cat has a face for radio!!!!!
You bet. :) Kitty was a one-eyed love machine that added to my understanding of life. May she rest in peace. :angel:

Listening Tests
First off, I think it is important to detail what standards I am testing against. I am testing against almost 18 years of uninterrupted, unmodded listening on this system.
I am using old and new source material that is familiar to me on my stock original system.
I did not 'burn-in' any of the caps before testing.

Configuration #1 -
Original config from Hafler =

50V 22uF BiPolar IllinoisCapacitor brand circa '86 bypassed with a rolled, .01uF 100V M. Polypropylene (SEACOR, I think).
Let us call this the reference standard.

Configuration #2 -
50V 22uF XICON NPRL series w/original bypasses.

I listened to these for quite awhile. The sound could be described as neutral compared to the IC caps. The bass was not as pronounced, but there was a nice, clean quality to the high end. Not as dynamic because of the slightly reduced bass, and lacking the 3D image of the original caps.
Still, I found these caps listenable.

Configuration # 3 -
(5) 4.7uF (24.2uF) 50V WIMA MKS2 in parallel with no bypass.

These are the 'supercaps' in the pics. I wanted to like this. It just seemed cool to go from electrolytic to all film across the signal path.
But...
I noticed a 'flatness' to the sound almost immediately. Disappointed, I listened to quite a few tunes, but there was a most peculiar trait - instead of depth, things in the rear of the soundstage simply sounded quiet not further away.
Bass was neutral, no better than the XICONs.
There was a lack of holographic effect and some of it seemed to come from a 'dead' sounding top end, this led me to my next config.

Configuration #4 -
Above config bypassed with .1uF 160V EPCOS MPP

Little change from the above. Disappointing.

Configuration #5 -
(4) 4.7uF (about 19.6uF) 50V WIMA MKS2 in parallel w/ no bypass.

Just clipped off the top cap and it's epcos bypass for this arrangement...
Better. A more open top end... but still no bass. I tolerated this arrangement for a night of listening, but I eventually began to get that yearn to hear that special something that only audiophiles understand.

So...

After all of the above listening tests were concluded over a couple months (mostly with the XICONs), I soldered the original IC caps back in along with their original bypasses. Beautiful, original stock - I even installed them to original channels and facing the original polarity.

WOW. It's almost sad. There is no comparison, it's like my system has come alive again. So much soundstage, so much bass, and ooooh so real.

All I can say is I feel extraordinarily priveleged to have owned this preamp in it's factory original condition. I have replaced the PS caps, but there is no comparison, the coupling caps at the output make all the difference. I noticed changes with the PS caps, but without the original IC coupling caps, this pre was dead in the water either way.

So, I guess it just goes to show that there are an amazing number of variables here - all of which can affect the sound.
I doubt I can ever best the original caps, so I sure hope they last a good long time.
 

Attachments

  • wimaepcos.jpg
    wimaepcos.jpg
    79.4 KB · Views: 336