EAR834 - component questions...

[netphreak]

Alright... many builders use 0.1uF (probably because a lot of PCB's specify it, as on mine), some use 0.012uF or 0.015uF...
Well... can I conclude that one may try anything from 0.01uF to 0.1uF (paper in oil is a good choice), and it will give small variations in "feel"?

Don't take me wrong - valuable information it is

[Jessica_K]

The original EAR had 12nf. And use 110pf in the RIAA feedback
Again a lot of changes are made here. Like 100nf an reduced to 100pf in feedback to sound better. So it is just personal preference

I would like to try a few more mods myself, but my TT is sick and quite costly to have it functional again so I'm limited to just working on paper while I save

[tubeactive]

  OK, I decided to use some arithmetic to decipher the variances in parts values. It is no surprise the original used 12nf (.012uf).
  Using a .1 uf (100nf) with the next stage 3.3Meg grid resistor yields a very low -3 db point:
 R x C with .1 uf x 3.3Meg =330,000 uSec =  -3 db at .483 Hz.
 Using a .01uf yields 33000 uSec = -3 db at 4.83 Hz.
 Using a .012, 39,600 uSec. = -3 db at 4.1 Hz.
 Using a .015, 49,500 uSec. = -3 db at 3.22 Hz.

  The lower values, .01, .012 or .015 uf are very sufficient for very high fidelity results. Moreover, these values are less likely to accentuate or resonate with any subsonic problems arising from mismatched arm/cart combos. While there are folks who will prefer the .1 uf, the original design value of .01 uf was sensible.

  When I look at the phono EQ feedback circuitry, the first value I notice is the 330 pf "turnover" bass EQ cap. This value is unique to the EAR phono preamp(s). It is very close, in numerical value, to the actual RIAA specified turnover time constant of 318 uSec = 500 Hz. With the only resistor within the feedback circuit of 750,000 Ohms, why did some versions choose 110 pf, instead of 100 pf, for the treble rolloff cap ? 750,000 x .000100 = 75 uSec, "proper" RIAA spec rolloff EQ time constant. 750K x .000120 = 82.5 uSec. = a bit high, according to RIAA specs, which can cause a "mellowing" of the treble.
 
 The June 1979 issue of the Journal of the AES includes the seminal Stanley Lipshitz compendium of RIAA EQ circuits, both active and passive. Typically, and almost universally with active feedback EQ circuits, there are two R-C networks in series. One network is the Bass Boost EQ (3180 uSec) network, the other the Treble Rolloff EQ (75 uSec) network. The series arrangement of these two (R-C in parallel connected) networks "interacts" and "derives" the third needed time constant of close to 318 uSec for the RIAA specified 500 Hz Turnover. That important formula is (R1 x R2 divided by R1 + R2) x (C1 + C2) = Time Constant in uSec. 159,155 divided by that time constant = Turnover Freq. in Hz.

  Trying to understand the math, can get confusing. RIAA specifies 3180 uSec (50 Hz) for the Bass Boost EQ time constant. However, EAR's honorable Tim d' P. utilizes a unique way to arrive at the required 318 uSec time constant specified for the 500 Hz turnover frequency. Since there is only an R1, we need to "substitute" an appropriate numerical value as a "place holder" for the "Open Load" for R2 which "would be" across the 330 pf Turnover/Bass EQ cap in a more typical active feedback EQ circuit.

  Bear with me, please, as this will make sense rather quickly. If you have a good digital VOM, like a Fluke, what value in Ohms would you consider an Open Load or near infinity ? When I check the charging speed of very good, low value coupling caps, say under .15 uf, my Fluke testing resistance across the cap, climbs quickly and flashes "OL" at around 24-30 Meg Ohms. That's a pretty high Ohmic value, yes ? Let us use 30Meg Ohms in the above formula, as the "imaginary" R2 across the 330 pf cap C2.
   (30M x 750K) divided by (30M + 750K) = 731,707.32, the multiplier. (C1 + C2) = .000100 + .000330 = .000430 uf.
   731,707.32 x .000430 = 314.634 uSec ! 159,155 (the mathematical constant) divided by 314.634 = 505.8 Hz.  That is pretty close to the 318 uSec or 500 Hz which the RIAA specifies. Even using 10Meg Ohms as the "imaginary" R2 across the 330 pf C2 in the formula, we would arrive at 300 uSec or 530 Hz, which is also very usable. As long as the "derived" turnover frequency is 500 Hz or a bit higher, we are fine. Too high in frequency (low number in uSec.) and the extra mids amplified could be heard, and easily seen on a response plot.

  So, with this ingenious way EAR designed this active feedback Phono EQ, the close to 500 Hz turnover is at +3 db above the typical 1000 Hz 0 db reference frequency and the playback curve's bass boost rises at it's natural and required 6 db/octave slope toward the lowest bass. However, since there was no Resistor across the 330 pf cap, the rising slope continues to subsonic frequencies.  That can possibly generate rumble or subsonic resonances, like woofer pumping, unless the design engineer limits the subsonic pass frequency with a built-in High Pass Filter. This subsonic limiting was accomplished using the original design's .012 uf coupling cap with the next stage's high value grid resistor !

  See ? I told you this would make sense rather quickly, right ? So, build your favorite version and enjoy ! Parts values and types will be like the colors on your artistic pallet, brushed on and used within the preamp, like your artistry.

[netphreak]

Oh dear... tubeactive - that was an overwhelming post! Been reading it a few times, will read it many more. I greatly appreciate your effort and time to give amateurs (me) a glimpse of how much there is to learn to become an audio engineer

What comforts me, is that it seems possible to replace components - listen, and draw conclusions. In other words - you don't *have* to understand it all, but yet improve the sound quality.

[needlekiller]

What comforts me, is that it seems possible to replace components - listen, and draw conclusions. In other words - you don't *have* to understand it all, but yet improve the sound quality.

no, you must understand the values of your coupling-cap and the resistance in the following stage!
so have a look here:http://www.sengpielaudio.com/calculator-RCpad.htm
hope this helps!

[netphreak]

Well, it seems there's plenty of room when it boils down to capacitors anyway. 22, 33, 47 uF - doesn't matter much. 0.1, 0.01, 0.12 uF - a different "feel" to the sound.

Almost none of the components delivered by Douk was equal to what's written on PCB, and what written on PCB doesn't equal the original EAR scheme.

Edit 1: I think I understand 0.1 vs 0.01 uF now... The next stage resistor must be 100000 when using a 0.01uF caps, instead of 10000 when using 0.1uF caps!
Edit 2: Don't read edit 1...

[netphreak]

Sorry to ask this questions which seems obvious to everyone... But where's the "next stage" resistor? From what tubeactive writes, it's the 3.3M resistor. But what about the 2.0M resistor? It's not part of the "next stage"? And if I replace the 0.1uF caps with 0.01uF caps - only the 3.3M resistor needs to be replaced as well? Just asking trying to understand...

[needlekiller]

in my early days i was one of the "bigger is better" guys but after diggin deeper i realeasd, thats not the right way! better is use the given values and use so called better components. like modern resistors or new caps. but in the most , i must say, the magic has gone! in the psu feel free, but in the rest of the amp dont change all at begining, go step be step and if a change of a res or a cap sounds better for you  use it. but dont think the result rules for all! i build my amps for my personal specs, no matter if it in any given hifi specs! the only result is when my friends swing there foots and say can you play it a little bit louder.
at the moment i listen with a el84 se in pentode mode and coral bb`s with aiwa magnetostas for highs in a diy solo-vox cabinett from auditorium23

[tubeactive]

 Hi netphreak, Fret not; You are on the right track. You are correct concerning "building a DIY kit versus full understanding of electronics." It is also worthy to learn more and more about circuitry, in order to gather and strengthen awareness. Knowledge becomes valuable as you will be able to build on that knowledge with what you have already learned, or remembered. Theory is tedious, though. Simply remembering some theory is all that is needed, until full understanding later "lights the bulbs" inside our heads. All in good time. For now, work safely and prepare yourself for ease of changing parts later. One of the builders posting in the other threads, Jay has built a beautiful preamp, which he has already torn down and rebuilt again, in search of finding the "slight hum" culprit. Arranging the parts with this potential upgrading in mind will help later.

   There are currently four threads about the 834 Phono stages on LH. I posted the math verified circuit info for everyone. My intention is certainly to enlighten, plus not confuse anyone. Let's see if this posted scheme from another DIY EAR 834 thread's (page 26) can help you and/or others understand some more:

              

  I also recommend you read and reread some of the rekinchin posts on these EAR 834P preamp threads. He is very knowledgeable and makes excellent recommendations concerning the various builds regarding this DIY project. You might want to print out this particular scheme or copy/save and then print, for it is a variation which may come in handily while deciphering your build. Note first that this scheme chooses two lower gain tubes than the usual 3 x ECC83/12AX7. 12AT7/ECC81 could also be another tube choice, as well as the 12AY7.

  Note the 150 nf/.15 uf with the next stage's 2 Meg grid resistor. R x C = 300,000 uSec = - 3 db at .53 Hz. That is a very low frequency, which could cause some rumble, as previously mentioned. A lower value coupling cap, like .02 uf would work fine, with a -3db point of 3.97 Hz. A .025 or .033 would also be fine. For some reason, .033 couplers are very reasonably priced, at least stateside. Demand must be low for that value.
  Now, observe the 12 nf/.012 uf across that 2 Meg, with the 3 Meg next stage's grid resistor. The original circuit intended both resistors. If you later change coupling caps, I recommend keeping the kit value resistors, at first. That last stage is the cathode follower, which inherently has a very high input impedance combined with a very low output impedance. Hence, the high 3 Meg input R to the cathode follower. FYI, the very low output impedance of a cathode follower enables two functions in this preamp circuit. This provides easy "load matching" for the next stage, a control unit or integrated amplifier, plus the cathode follower also easily "drives" the phono EQ (active feedback circuit). With a little more electronic awareness and parts juggling, changing that 68K cathode load resistor to about 100K would enable the .012uf/2 Meg to be bypassed (jumpered across with wire), thus enabling "direct coupling" from the previous stage, as rekinchin has posted in the larger thread. Sonic benefits would definitely result. You can do this later, after more understanding down the road or after you have completed the kit version first.
  Simply following the kit instructions and the posts in the other threads will lead to success. Changing cap types and values will still enlighten your ears, remembering to leave break-in time before presuming any subjective results with parts swapping. I recommend only changing one cap value per attempt, thus avoiding indecision. Some folks only change one channel's single cap at each change attempt. Do what you wish. This is your build...I believe the final coupling cap, the 1 uf from the cathode follower, will be essential for you to critically choose which cap type "helps" the sound, (for example polypropylene vs. PIO vs. paper/mylar/polyester vs. polycarbonate vs. teflon...). You may wish to post your finalized coupling cap decisions, as many of the builders might benefit from another set of discerning ears' choices...ENJOY !

[Jessica_K]

I have been playing on simulation with some interesting tweaks to values. My theory is make it flat 20-20khz and "cough" have a low rumble TT. Linn have a phono amp that is spec flat to +/- 0.2db over that range. Analog filters can't just cliff after 20hz so the best 3db point is likely to be about 2-4hz unless they are throwing opamps at like confetti with high pole filters.

Well I do have some values and tweaks up my sleeve that can give a flat response to +/-0.1 worst case with 1% components and is based on the fact that other impedances in the design and miller effects are not taken into account with the "simple" RIAA component calculations and so the values need to be modified to take all in.

I have not tested any of the theory in a real amp and I expect it will depend a lot on the tubes used

I'm currently building a RIAA curve source to test the current build I have first. Then try those values out and roll a few tubes and see what difference it makes

This is quite academic with the only emphasis being to get a repeatable flat response. I have no idea what it may sound like or if indeed I will need a rumble filter after all but it will be flat down to some defined frequency

Robert

[Jessica_K]

Hi netphreak

Don't change the 3.3meg and 2meg resistors as they set the DC bias for the output cathode follower stage. The cap acts as a high pass filter with the 3.3meg as the 2meg can be kinda ignored for AC (sound) so changing the caps value effects the low frequency roll off.

There are mods that can be made in this area like removing all 3 components and just linking but care needs to taken to ensure the DC bias is set valid as wrong setting could cause much larger currents to flow in that last stage than the valve can take or even to make the grid positive with regards to the cathode ( a big no no )

Robert

[netphreak]

750,000 x .000100 = 75 uSec, "proper" RIAA spec rolloff EQ time constant. 750K x .000120 = 82.5 uSec. = a bit high, according to RIAA specs, which can cause a "mellowing" of the treble.

Trying to follow: Is 0.000120 correct? Are you referring to the 110pF caps, and it should be 0.000110? Assuming this, here's some math:

R1: 2M resistor + 3.3M resistor = 5.3M resistanse (5 300 000)
R2: Feedback resistor = 750K (750 000)
C1: 330pF (0.000330)
C2: 100pF (0.000100)

Formula: (R1 x R2) / (R1 + R2) x (C1 + C2) = Time Constant in uSec

(5 300 000 x 750 000) / (5 300 000 + 750 000) x (0.000330 + 0.000100) = 318

Now, I received and soldered on 110nF caps for C2:

(5 300 000 x 750 000) / (5 300 000 + 750 000) x (0.000330 + 0.000110) = 325.3953488

Why should I not replace my C2 with 100nF caps, if what we're after is 318uSec (500Hz)?

[Jessica_K]

Hi Netphreak,

I agree with how the math works but following my previous post, the math assumes that there is nothing driving the circuit and nothing taking any output from the circuit. All the other components have an effect on the impedance that shift the time constants by small amounts.

below are some simulations of the response for different values of components in the schematic


Using all original values for C7 (110pf) C4 (10nf), the response is +/-0.475 and has a bass peak around 30Hz and rolls off from 1Khz, some say this sounds dull

Using values for C7 (100pf) C4 (10nf), the response is +/-0.375 still has a bass peak around 30Hz but now apart from a dip about 300Hz has improved high frequency response sounding brighter

Using values for C7 (110pf) C4 (100nf), the response is +/-0.448 and has a extended bass beyond 20Hz but still rolls off from 1Khz, sounding dull but with a deeper base (possible rumble too)

Finally Using values for C7 (100pf) C4 (100nf), the response is +/-0.354 has a extended bass beyond 20Hz but now apart from a dip about 250Hz has improved high frequency response sounding brighter


So you can see none of these values are perfect and each provide a reasonably "flat" response in the greater scheme, but each will have subtle sound difference
below is the greater scheme response curve 1Hz to 100KHz for the original component values and using 100pf at C7


Robert

[netphreak]

Most of my chosen components have now arrived, and building has started. I bought an r-core transformer with the specifications (secondary):

240VAC @ 0.05A
6.3VAC @ 2.0A
3 x 9.0VAC @ 0.2A (will be left unused)

When measuring the output voltage, it reads 285VAC and 7.4VAC. Now, of course - there's no load present - so this is kind of expected, right? Well, I connected the PSU board, and it doesn't lower the voltage. Should I be concerned connecting the preamp/RIAA to the PSU board?

edit: exact measurements of the DC side of the PSU: 300VDC and 9.8VDC with no load. This can't be "by design"??? Mains measured to 228VAC.

[netphreak]

Had to try in the end... With both PCB boards connected (including tubes), I now have 6.5VDC & 299VDC. Still a bit high, but maybe this will drop further when adding audio in/out and actually play some music?