Eros Alternate Build Thread

[xcortes]

I have been doing some research on upgrading parts for the Eros and instead of taking notes on a paper I decided to post them here for future reference. Moreover, since I will actually build an Eros with important changes I will document the build here.

All of the changes will have to do with parts and construction. I have no plans to modify or alter PJ's circuit whatsoever.

Of course the ultimate upgrade would be to buy a Repro. I have a tape Repro and it is just fantastic. But some of us are living proof of Beranek's law and we like better our own built gear.

My idea is to use premium parts where it makes sense. I don't have a specific budget but I think it would be adequate to limit the cost to the same cost as the kit (including the chassis but not considering tubes).

For tubes research please refer to Ironbut's research documented on the Tape Project Forum. The thread is for the Repro but the circuit is pretty similar so his observations should apply.

So here we go.

Let's start with the chassis. I will build a Front Panel Express chassis. FPE is very user friendly. Once the design is ready I will share the files here. The chassis will have a 19 inch rack front panel since I want it to match with my Repro. It doesn't have to be as shallow as the Repro, though. And frankly fitting an amp in such a small cabinet is a gigantic task (hats off to Doc and PB here). I also don't want to do mayor mods to the layout.

The layout will be similar to the original except that it wil be in a longer chassis (the back panel) and will be on the vertical plane. However the main amp and RIAA sections will be on the horizontal plane where there's lots of real state. One thing I don't like about the Eros is that the C4S boards are above the circuit. Not a problem if you get it right the first time but a pita for troubleshooting.

I hope all this is much clearer when I post pics of the actual chassis.

I also plan to include two sockets for SUTs on the back panel.

Please feel free to comment as I document my findings and construction over the next months.

Next: Capacitors Part One

 

[xcortes]

Ed Fallon on the old forum:

If I was to upgrade caps in the Eros kit, I would only do the output coupling caps first. I like .47uF 400V Auricaps in the prototype and they sound great. The kit will likely come with .47uF 400V Solens which aren't too bad either. The RIAA caps and two cathode resistor bypass caps are carefully selected and don't merit changing without very careful consideration.

I have always liked Auricaps and I will definitely replace the output caps with Auricaps (note that the Eros stock output caps are 1.0) not 0.47. In general, 0.47 should work.

PJ:

The output stage of the Eros is the same 6922 tube as Seduction, so Seduction's 0.47uF should be adequate. The stock Eros output is 1.0uF Solen

But since I plan to run a preamp with a reflected impedance of around 25k (a 1k pot after 15k:600 transformers) I will use 1.0uF.

PJ again:

when we developed the "Sweetest Whispers" we made up a passive preamp using that 15K control as a conventional potentiometer. A Seduction could drive it as is (I would probably increase the output capacitor from 0.47uF to 1.0uF myself, but it works OK as is)

 

[xcortes]

PJ on the old forum:

The RIAA resistors are 1% and the capacitors are 3%, so greater precision would start with locating 1% capacitors. They do exist (or can be selected if you have a precision capacitance meter) - they are simply too expensive to include in a kit of moderate cost. Incidentally, the values (0.010uF and 0.030uF) are chosen so that one can buy 8 identical 0.010uF capacitors and parallel three of them for each 0.030 value. Often one can get a better price on 10 caps. By eliminating the two least accurate ones, choosing the best matched pair for the 0.010 values, and choosing the threesomes carefully, it should be possible to get some very good matches with 3% or even 5% caps - and exceptional performance from 1% caps. The exact ideal for the smaller cap is 0.010288uF, but the 6922 Miller capacitance supplies about 50pF and there is a few pF of stray capacitance, so 0.0102uF would be ideal. IMHO, matching between channels is more important that getting the corner frequencies exact

For the 0.03 I will parallel Auricap 0.015 caps. These are 10% but some sellers match them for a fee. A 0.015 600V Auricap at Parts Connexion (PCx for the rest of the thread) costs $11.02 and they match a pair for $1.

For the 0.01 I will use V-Caps ($39.36 each from VH Audio when you buy two). At that price I would expect them to sell a matched pair but I need to ask. I haven't found a published tolerance on these.

 

[Grainger49]

Xavier,

The sellers that match will give you two of a matched value.  That value wouldn't necessarily be the value on the cap.  For instance a 0.015uF 10% cap might give you a matched pair as far off as 0.0135uF.  That is less accurate.  I went on a search for these values at the big suppliers in Polypropylene caps and didn't find much that was 1% or 2%. Nothing the right value.

 

[xcortes]

Thanks for clarifying that Grainger.

Probably better to buy eight 0.01s and match them as PJ suggests.

I did find now that V-Caps are 5%. PCx sells the 0.01uF for $35.42 when buying five or more.

Auricap does make 0.01uF 1% ones but is the tined copper rdial leads version and I haven't found a source.

 

[Hank Murrow]

Dear Xavier;

You might want to study the Eros custom-build that Paul Birkeland did so beautifully for me. I finally got a pair of the Altec-Peerless 4722's for Lo-output MC's. Anyway, here are the links for the Phonostage..... and Godspeed to you!

http://www.bottlehead.com/smf/index.php/topic,621.0.html

http://db.audioasylum.com/mhtml/m.html?forum=magnequest&n=9737&highlight=Hank+Murrow&r=&search_url=%2Fcgi%2Fsearch.mpl%3Fsearchtext%3DGyro%2Bse%26b%3DAND%26topic%3D%26topics_only%3DN%26author%3Dfretless%26date1%3D%26date2%3D%26slowmessage%3D%26sort%3Ddate%26sortOrder%3DDESC%26forum%3Dvinyl

http://db.audioasylum.com/mhtml/m.html?forum=magnequest&n=9738&highlight=Hank+Murrow&search_url=%2Fcgi%2Fsearch.mpl%3Fsearchtext%3DGyro%2Bse%26b%3DAND%26topic%3D%26topics_only%3DN%26author%3Dfretless%26date1%3D%26date2%3D%26slowmessage%3D%26sort%3Ddate%26sortOrder%3DDESC%26forum%3Dvinyl

Cheers, Hank

 

[Paul Joppa]

One of the differences between Eros and the Repro is the servo bias, which requires a large, low-voltage electrolytic cap at the EF86 cathode. We chose the OsCon or equivalent (organic semiconducting electrolyte) which has exceptional ESR over a wide frequency range, as well as a good reputation. But Doc B says the LED in the Repro does sound a tiny bit better in a really high resolution system - and he has the most experience with both. Unfortunately, most EF86s won't bias up right; only the best NOS ones seem to work and they are pretty expensive when you can find them at all.

So my take is, see if you can find a better capacitor. The stock value is 2200uF/4v, but I think it does not need to be precise - 1500uF or greater, and 2v or greater, should work. Some things to watch for:

1) The actual voltage is low, 1 to 1.5v in operation usually. At this low a voltage, forming and maintaining the insulating oxide layer inside the capacitor becomes marginal, and some combinations of metal (usually aluminum) and electrolyte (there are several) may not work as well as others.

2) The capacitor is in a very sensitive spot in the circuit, and a physically large capacitor would be susceptible to noise, picking up all kinds of interfering signals like an antenna.

3) The organics are leaky, two or three times as much leakage current as standard electrolytics. That's enough to carry nearly half the current in this spot. If it's much leakier, the servo will be effectively shorted out, so check the leakage specification.

For what it's worth, I have recently heard good reports of the "Siltec" capacitors, but have not yet heard them.

 

[xcortes]

Thanks Paul.

I was about to start researching on those cathode caps.

There's a Black Gate NX 2200μF / 6.3V NX. Not that large at 18mm x 35mm and within the enclosed cabinet maybe won't pick up noise. Of course I don't know if it will maintain the oxide layer at 1V. There's only ne way to know I guess.

How about the 100UF 160 V bypass cap?

 

[Grainger49]

IIRC, the Eros was first developed with a 100uF@160V for the other cathode bypass cap too.

PJ posted about these caps on the old board and I believe I copied it over to here.

Yup!  Here it is:

http://www.bottlehead.com/smf/index.php/topic,410.0.html

Here you can see it was Paul who told me it was a BG cap but in the Repro.

And are you up early or late?

 

[xcortes]

Thanks Grainger.

Actually I was early!

Hmmm, so according to PJ:

Generally electrolytic capacitors work best at 30% to 80% of their rated maximum voltage.

So for a 1 v voltage we need a cap rated at 1.5 to around 3v. So the BGs are huge at 6.3v and may not work.

And more info. This time by Ed Fallon:

The polymer electrolytic parts were selected for compact size, price, spec, performance and availability.  If you can find another cap that fits and performs well, go for it.  The parts we chose were an attempt to mimic the well regarded Sanyo Os-con.  I would recommned the Sanyo to swap if you can find them in stock somewhere.  

I found a 1,500 uF:

http://cgi.ebay.com/4-Sanyo-SVPC-2-5V-1500UF-SMD-Aluminum-solid-Capacitor-/300354916191?pt=LH_DefaultDomain_0&hash=item45ee8c4f5f

Although this SOB ships "worldwide except Mexico"!

Maximum leakage current is specified as 750

 

[Grainger49]

The cap Bottlehead uses in that position is a 4V cap.  The ones I saw were OsCon, the organic semiconducting electrolyte.  Possibly the fact that they are leaky allows for it being used under 30% of the rating.

I would bet that a BG would sound fine in that position if you can find it.  They have become quite scarce and the prices are through the roof.

I have seen that post by Ed but couldn't find it to bring over to the new board.

You have a PM.

Seller says sent.

 

[xcortes]

OK, so the OscCons are purchased. Thanks Grainger!

I have already decided on a switch. It's an NKK that looks very similar to the one on the Repro. Remember that I want this unit to look as similar as possible to the Repro. I'll post the part number later. It is important to use the same switch (or one with the same thread) because it will be threaded onto the front panel.

I also decided to mount the SUTs inside the amp on the opposite corner to the PT. It was tempting to installe the sockets on the back but these guys are very succeptible to hum pickup so I think it's better to have them housed inside the amp. I also decided not to install a MC/MM switch. Neither a gain switch for the SUTs. I will mount them using one of these:

http://www.partsconnexion.com/product9751.html

And I intend to use the same strips:

http://www.partsconnexion.com/product3172.html

I made lots of headways yesterday with the chassis design. In order to finalize it I need to finish planning the layout. That will take me somedays.

Next, 100uF capacitor.

 

[xcortes]

PJ on the old forum:

The 0.1uF on the shunt reg board is for stability at ultrasonic frequencies. It is not thought to be a player in the sonics of the amp.

Of course, the reason for using a direct coupled circuit is to eliminate the interstage capacitor, which leaves only the output capacitor in the "signal path".

Nevertheless, the cathode bypass capacitors do carry signal current and are candidates for replacement. That would be the 2200uF/4v and the 100uF/160v caps on the PC boards.

I quoted the whole sentence because I was wondering about the 0.1uF ones on the SR board. The stock ones will stay.

But how about the 100uF/160v. Seems like a candidate for replacement. In addition to the natural Black Gate candidate what would be other possibilites? How about non lytics?

 

[Grainger49]

Yes, that is the one I meant in Reply #8 above.  The electrolytic cap can be replaced with a non-polarized electrolytic or a better, say Black Gate (even polarized BG is better).  If you can find them.  Michael Percy lists one that value and voltage.

 

[xcortes]

Would a 150uF one work in this position instead of the 100uF?

 

[Paul Joppa]

1) Yes, I believe the 100uF bypass across the Eros 6922 cathode resistor is a player, and a better capacitor is likely to make an improvement.

2) The 100uF and 2200uF cathode bypass capacitors are chosen to maintain suitable phase margins for stability in the servo bias loop. The symptom would be a very slow pumping action, which should be easily audible. But so far in my experimentation, I have never actually produced this problem. My analysis and modeling is too crude to give detailed answers (also the tube and transistor gains will vary between devices, and with temperature) so I can't say clearly what values will not work. For what it's worth, making the small one larger and the large one smaller would be going in the wrong direction.

 

[xcortes]

Thanks Paul,

that's good info.

 

[Grainger49]

  .  .  .  For what it's worth, making the small one larger and the large one smaller would be going in the wrong direction.

So the 100uF could be smaller and the 2200uF could be larger?  Or are you referring to the voltage rating or physical size?  That would make my above statement vice versa.

 

[xcortes]

I have the layout ready. The section above the dashed line is the vertical plane on the back. The section below is the horizontal plane on the bottom of the chassis. I think I'm ready to start detailing the FPE chassis.

 

[Paul Joppa]

So the 100uF could be smaller and the 2200uF could be larger?  Or are you referring to the voltage rating or physical size? 

Haha! Sorry, it was unclear and has been a confusion before. Yes, from a stability standpoint the 100uF could be a lower capacitance and the 2200uF could be a larger capacitance. Below about 50uF, the 6922 cathode bypass will have an audible effect on bass extension and I do not recommend it.

I think 160v is close to an optimum rating for the 100uF if it is electrolytic, and the 2200uF should have a low voltage rating, like 4 or 6.3 volts. Higher voltage ratings should work; they will eventually lose their ability to withstand the full rated voltage but should continue to perform well at lower voltages as long as it's enough to maintain the chemistry. Sorry I don't understand that low-voltage chemistry very well yet.