Separate power supply vs. single chassis

[tube fanatic]

Just an informal survey:  would you prefer a single chassis amp to one with a separate power supply, or would it make no difference to you?  Reasons for either view would be appreciated.

There's no question that isolating an amplifier's power supply to its own chassis, and locating it a distance away from the amp itself, results in a superior sounding unit.  It has always been my preferred method.  In spite of that, on occasion I'm asked to use a single chassis even if it results in a bit more noise, powerline grunge, and so on.

Looking forward to comments!

Maynard 

[Coytee]

I'm ignorant enough that it doesn't matter to me.  If those that know say it's better then I'm open to it.  If they say it's no difference then I can accept that as well.

 

What I would have wanted to see is an amp (mono or stereo) that has XLR inputs and might have enough oomph to do HT front speaker duty. 

 

(I actively biamp and my active has XLR outputs)

[JJkizak]

The best regulated power supplies I have seen were the old Western Electric 5 and 6 tubed paralleled regulators in the Mk 25 Mod 3 fire control radar. You could vary the input voltage from 130vac to 90vac with the output voltage remaining rock solid (output voltage movement imperceptible) I suppose if your looking for WWIII performance it would be nice to separate the power supplies from everything else.

JJK

[Schu]

mono's from here on out for me.

[John Chi-town]

I'm ignorant enough that it doesn't matter to me.  If those that know say it's better then I'm open to it.  If they say it's no difference then I can accept that as well.

 

What I would have wanted to see is an amp (mono or stereo) that has XLR inputs and might have enough oomph to do HT front speaker duty. 

 

(I actively biamp and my active has XLR outputs)

 

Adcom has the vintage 5802, Nelson Pass designed with XLR's.  In addition their new offerings like the 555SEII, as well as the 565SE both have XLR inputs.  I own the 565SE rated at 250 WPC.  I bench tested to 310 watts prior to Instantaneous Distortion Lights comeing on.  More than enough power for what you are looking for, especially with Klipsch speakers.

 

Best regards,

John

Edited August 29, 2014 by John Chi-town

[Rick J B]

Interesting question.

 

I'm in the process of gathering parts to build a Tubelab simple SE amp and have been seriously considering a seperate power supply. From what I can tell there doesn't seem to be a down side and anything that might make it quieter is a win IMO.

[Fjd]

Maynard, is it safe to assume you mean tube amplifiers and you have not branched into solid state amplifier builds?  I’ve seen people with separate power supplies for solid state amplifiers and end up stacking everything; however, I suspect that if the transformers are toroidal or a well-shielded E-I core, stacking may be a non-issue.

 

From an amplifier perspective, I remember seeing examples of the old ham radios and several models of the Heathkit amplifiers using the separate PSU chassis approach.  There seem to be a lot of more recent examples of the two-chassis pre amplifiers separating the PSU such as the audible illusions m3a and the old Audio Research pre amps.

 

For me, impact on sound quality aside for a moment, from a basic user perspective, sometimes the convenience of everything in one chassis is nice from taking up less ‘real estate;’ however, when the weight of a tube amp starts to climb, the separate chassis approach may end up being more convenient for other reasons such as being too heavy to move around easily.

 

Since I am such a novice when it comes to DIY amplifier building, out of curiosity;

 

What are you using for an umbilical cord and connectors between the PSU chassis and the signal chassis?

 

How are you handling filter caps, chokes and rectification?  For example, is rectification in the PSU chassis where you are carrying all of the DC in the umbilical cord or in the signal chassis where you would be carrying all of the AC in the umbilical cord?

 

How are you handling grounding? For example, a star ground in the PSU chassis and a ground through the umbilical cord to connect each chassis?

 

[tube fanatic]

Maynard, is it safe to assume you mean tube amplifiers and you have not branched into solid state amplifier builds?  I’ve seen people with separate power supplies for solid state amplifiers and end up stacking everything; however, I suspect that if the transformers are toroidal or a well-shielded E-I core, stacking may be a non-issue.

 

From an amplifier perspective, I remember seeing examples of the old ham radios and several models of the Heathkit amplifiers using the separate PSU chassis approach.  There seem to be a lot of more recent examples of the two-chassis pre amplifiers separating the PSU such as the audible illusions m3a and the old Audio Research pre amps.

 

For me, impact on sound quality aside for a moment, from a basic user perspective, sometimes the convenience of everything in one chassis is nice from taking up less ‘real estate;’ however, when the weight of a tube amp starts to climb, the separate chassis approach may end up being more convenient for other reasons such as being too heavy to move around easily.

 

Since I am such a novice when it comes to DIY amplifier building, out of curiosity;

 

What are you using for an umbilical cord and connectors between the PSU chassis and the signal chassis?

 

How are you handling filter caps, chokes and rectification?  For example, is rectification in the PSU chassis where you are carrying all of the DC in the umbilical cord or in the signal chassis where you would be carrying all of the AC in the umbilical cord?

 

How are you handling grounding? For example, a star ground in the PSU chassis and a ground through the umbilical cord to connect each chassis?

Excellent questions!  I should have elaborated a bit more in posting.  Yes, I'm speaking of tube amps exclusively.  Your point about the single chassis requiring less space is what is usually cited even though the chassis is larger.  Weight isn't a factor in the amps that I'm considering as they are low/flea power SETs or SEPs with a power xfmr which doesn't weigh very much.  Depending on the design, the umbilical can be either a few feet of #16 or #18 lamp cord, or up to 8 individual (but tied) #20 wires.  For the former I've used a nice Cinch Jones plug on the lead from the amp, and the latter an octal (identical to a tube base) plug.  However, in both cases I'm switching over to Anderson Powerpoles due to the presence of high voltage on the plug contacts for up to 15 seconds after the equipment is shut off (for example, if one were to instantly unplug the amp from the p/s and touch the plug pins while standing barefoot on a concrete floor the results may not be as desired!).  Only the input filter cap is in the p/s, and all additional filtering is in the amp (which is why the plug pins can be "hot" immediately after shutting down- it takes a little bit of time for the caps to discharge through their bleeders).  In the designs using the lamp cord, very low ripple dc is fed to the amp for both the filament supply and high voltage.  With the multi-lead arrangement, the low ripple dc is brought in for the high voltage, but the filament supply is generally ac.  The p/s chassis is star grounded and is also grounded to the wall outlet through the green lead of the 3 conductor power cord.  And the amp is star grounded in its own chassis with a dedicated lead to the p/s which puts both chassis at ground potential.  No ground loops occur with this arrangement.

I posted this question as I'm on the fence as to how to set up a bunch of amps requested by my local "characters."  Most favor the 2 chassis arrangement.  All comments and ideas are welcome!

Maynard

[Thaddeus Smith]

I actually have mine separated, though it's mostly for practical purposes. I'm using an old HAM linear power supply that won't fit in my amp chassis and the Class D board I'm using can sometimes be quite bitchy when it comes to electrical noise - so it just makes sense to keep them separate.

[tube fanatic]

Thaddeus, what kind of voltage are you running in that class D amp that you need to use a p/s from a linear amplifier?  I've been involved with many hams who needed a separate p/s to provide voltages in the 3k range for their linears.  Of course, we're talking about rack mounted stuff like that shown in this video:

 

http://www.youtube.com/watch?v=n470MunpSKg

 

Nothing like mercury vapor rectifiers to add a nice visual experience (Steve Deckert must like a glow also as he's using gas regulators in his Mini Torii amplifier)!  In spite of the locking cabinet shown in that video, I sure don't like those exposed terminal blocks and high voltage standoffs!

 

Maynard

[Thaddeus Smith]

The operating voltage range is 4.5v to 26v, with optimum power output at about 21v. Astron bench top supplies are becoming a popular linear power solution for these amps and since I'm running such efficient speakers (Cornwalls) in a near field configuration I don't need to get anywhere close to the 50w rated output. I was asking my HAM head father in law about a source for Astrons (12v, varying amperages) and he dug out a crusty looking power supply that is rated at 12v and 7a. Looks old and ugly, but he put meters on it before handing over and everything tested solid. Can't beat free.

 

http://www.ti.com/product/tpa3116d2

[erik2A3]

Hi, Maynard

Certainly a worthwhile consideration. For my part, well designed single chassis. By well-designed, I am obviously referring to comprehensive isolation of low level signal and PSU components. While I think the use of a separate chassis for B+, filament, bias, etc. supplies can indeed be beneficial, the benefit is one that in my experience pertains primarily to headphone amplification. I corresponded with Moth Audio (now Eddie Current) several years ago about this very subject, and a number of their current (pardon the unavoidable pun here) products, which in fact are also intended for headphone use, do use an isolated power supply chassis. I also built the Moth Audio Si2A3 with on-board switching for either headphone or loudspeaker use, but found residual noise problematic for headphones. This particular amp is direct-coupled between input and output, and so is entirely without interstage coupling capacitors. It is a single-ended triode design, though well-filtered and regulated DC is used on all filaments, including the voltage gain stage. As you know, the common 300B, 2a3, and 45 triodes have directly heated cathodes. If one is using AC filaments, PSU isolation is not going to provide much of a benefit. A separate chassis 20 feet away would not ameliorate the 60Hz noise artifacts associated, and usually simply accepted, with this type of amplifier. However, I will say that it's possible to build a very, very quiet filamentary cathode amp. Grounding is critical, as you of course also know.

I am familiar with some very clever single-ended designs that make use of 25Hz filament supplies, which tackle the directly-heated AC filament problem in what in my view is a particularly intelligent way.

A single chassis for both signal and power supply can be very satisfactorily quiet. I have a number of amps that are, for all intents and purposes, very nearly dead-silent with our big and very efficient horns. I am thinking in particular about the LEAK ST20 I overhauled a couple of years ago, as well as one very fine little amp you mentioned above, the Decware Mini Torii. The latter is probably the only amplifier I own that I didn't build myself and have done nothing to it other than to add headphone capability. Outstanding performer!

I don't know, I guess I just have a preference for a single chassis -- shorter lead length, more straightforward implementation overall.

Erik Mandaville

[russ69]

If the power supply is in the chassis, the designer is in control of the problems created by the power supply. You can design it for the least amount of negative impact. The other thing to remember is that as soon as you make the electrical connection the power supply is no longer a separate item, it just has a longer cable run and an extra connector.

[tube fanatic]

Hi Erik, and thanks for the point of view (and my wife happens to agree with the single chassis approach even if it is somewhat larger!).  You are absolutely correct that if the filaments are run on AC, having the p/s isolated becomes less valuable (although it still makes some difference by keeping the powerline wiring farther away from the signal circuitry).  The true advantage is with my full DC amps where the ripple coming into the amp chassis is around .25-.3%.  Additionally, with these being budget style amps using open frame opts (Hammond 125CSE), there's the issue of magnetic field pickup from the power xfmr if on the same chassis.  When using voltage amp tubes like the 12AX7 (as opposed to metal types like the 6SF5), I've on occasion had to go with steel shields when running the gain on the high side to keep things as quiet as desired.  Some of the amp guys I know install a sheetmetal shield across one side of the chassis (underneath) to fully isolate the power supply wiring from the signal side, but that's more work than I want to do!   In any event, I'm being very "anal" about these designs as even those with the onboard p/s have only a trifling amount of hum audible with an ear against the grill cloth instead of being totally silent.  

Russ, I appreciate your thoughts as well, and for the most part agree except as it pertains to the full DC versions as described above.

I hope more comments are forthcoming.  These amps are going to try to satisfy requests for outstanding sound at very low cost.  The budget suggested to me was $200 using all new parts, but I think it's going to come in closer to $300 if I'm to avoid severe compromises in the opts.  With the Hammonds I can get the desired bandwidth, but haven't seen anything significantly less costly which can accomplish as much.

Thanks again guys!  

Maynard

[seti]

I've only just started tube DIY and I've made some tremendous sparks... I mean mistakes..

 

I'm working on the layout of a stereo 45/2a3 direct coupled on one chassis. There isn't a lot of info on layout. If anyone has any links or book suggestions I would appreciate it.

[tube fanatic]

Please keep the comments coming as I want to get a sense of what is the preferred format.  I found some pix in the file of representative examples of both formats and have attached them so they can be considered together.

 

Maynard

[erik2A3]

Maynard:

I agree with what you outlined as far as grounding -- which becomes all the more important with the increased complexity of a separate PS chassis. I have also used the sort of in situ sheilding you mentioned using sheet aluminum stock to separate the PSU from signal components -- that can be very effective. I have also used two separate chassis with the umbilical on the inside rather than on the back with panel mount make/female connectors.

Something that might help find the best solution, particulalrly if you are planning to build multiple editions of this component, would be to build it both ways in order to determine if THIS design is one that truly warrants the additional machining and complexity of separate chassis. Of course this is extra work, but it only needs to be done once. It would save considerable time down the road if you discover that, despite what some have said about preferring two-chassis over one, you found no remarkable difference in perceived and measured noise. In other words, perhaps consider a single chassis unless the use of two results in a truly marked improvement in power supply rejection. A single chassis would also save cost. Are enclosed transformers and chokes out of the question? You can, of course, always install separate filament transformers (my preference) inside the chassis, with the just PT and OPTs topside. My thinking is that, with a thoughtful layout and clean wiring (which I have always seen in your work!), you would be able to build a unit that was every bit as quiet as a two-chassis design. As we've discussed, grounding is a critical factor. You mention star grounding, which always helps. Where you put your ground nodes is as you know also important. Grounding the power transformer center tap at the very first filter capacitor has always proved effective. Preaching to the choir here....sorry about that!

Great to hear you are using bleeders in your power supply! Not all designers do, and as I'm sure we both know and have learned on more than one occasion, filter capacitors can store a very respectable kick! Good luck with your project!

[erik2A3]

So, it seems we were posting here about the same time! That setup does look good. So, are you sending unfiltered B+ to the amp chassis?

Edit. Ah ok....I see how you did it.

Edited August 31, 2014 by erik2A3

[erik2A3]

Seti:

There a number of sources available -- them following off the top of my head contains some excellent information on vacuum tube audio electronics: the section a grounding among the best I have seen on-line.

http://www.aikenamps.com/index.php/white-papers

[joessportster]

I prefer seperate Power Supply for quieter operation,,,,,,,,,,,,,,,,,,,But a good single case design can be plenty quiet.  I also prefer the look of seperate