LRiaa inductors

stephie's place

Moderator: sbench

Post Reply
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

LRiaa inductors

Post by dave slagle »

Pick a Choke.... any choke

i started this as a new topic since it could get ugly...

I have been scratching my head a bit lately about where to make the choice of compromise for the LRiaa inductors and came to the conclusion that i am clueless :-)

this is where you guys come in. I wound and tested 18 bobbins to cover the range and gapped them a bunch of different ways. all measurements were made with my B&K 885 at 1V. I'll attach comments to the images and feel this will be a two step process. The initial step is to get some chokes to SB based on the measurements, then let the c3 contingency attack the edges of reason.

i have a pair of all of the bobbins so the whole lot can be shipped out as a package that will be sure to drive people in high altitudes mad. I have a number of thoughts but i'll hold them back until you guys get a peek at the pictures. I do think the general approach will be to pick a bobbin that gives you just over the desired L (however you decide to measure it), gap it and get it stable to the slightly higher value (however you decide to do that) and then remove turns to nail your desired value. It is important to note that i only measured the .002 and .004 gaps and essentially the gap is infinitely adjustable.

dave
Attachments
this simply represents the ac flux levels at the various frequencies for each bobbin for 1V of signal.  Note that 1V is probably substantially higher that the actual signals these inductors will see.
this simply represents the ac flux levels at the various frequencies for each bobbin for 1V of signal. Note that 1V is probably substantially higher that the actual signals these inductors will see.
image001.gif (7.32 KiB) Viewed 17434 times
this is more of a data and proof image.  The DCR line simply shows the DCR of each bobbin and the other two sets of plots show the bobbin as an air core and with just the e's inserted (huge gap). Note the linearity of the air core!  The fact that all thre
this is more of a data and proof image. The DCR line simply shows the DCR of each bobbin and the other two sets of plots show the bobbin as an air core and with just the e's inserted (huge gap). Note the linearity of the air core! The fact that all thre
image001.gif (8.62 KiB) Viewed 17434 times
my guess is sb will request bobbins in the 27-29 range with large gaps.  (note the crossed lines and resonance) My hope is the c3 guys will try some of the options that look a little ugly but might teach us something.
my guess is sb will request bobbins in the 27-29 range with large gaps. (note the crossed lines and resonance) My hope is the c3 guys will try some of the options that look a little ugly but might teach us something.
image001.gif (14.17 KiB) Viewed 17434 times
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

I do think the general approach will be to pick a bobbin that gives you just over the desired L (however you decide to measure it), gap it and get it stable to the slightly higher value (however you decide to do that) and then remove turns to nail your desired value. It is important to note that i only measured the .002 and .004 gaps and essentially the gap is infinitely adjustable.
Actually, the problem may be simpler than that. With the LR, unlike the Tango or LCR system, you actually have an extra degree of freedom or two. With LCR, the good news is that so long as you terminate correctly, the source impedance is a don't care condition. This makes it "easy" to drive from a tube view, since the Zout variation from part to part even within the same type doesn't matter. However, the inductance value and the resistance of the part are critical.

With LR, the terminating and source impedance both matter, so you lose that one degree of freedom. For it, however, you can more easily scale the impedances and if you drive from a low Z (thus the step down transformers), the variation in Z becomes less critical.

That's a long way 'round of saying, that what we're looking for is REPEATIBILITY. The exact value can be almost anything, since we can scale the two R values per section very easily. The "1Hy" should be viewed as a ballpark. The EXACT value is totally uncritical but should be the same from part to part and over level and frequency. This is so that someone can duplicate the circuit and get substantially the same results without resorting to "tuning". Ditto the 1.8Hy.

On a related note, just to make sure we're all on the same page, is the inductor intended to be the same form factor as the "standard attenuator" iron? (that's what I have laid the PCB out for. I have not "fabbed" the PCB yet, so I can change if we need to.

Steve
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Ok so lets do the first go at it for linearity and as much as the values don't matter, i think sticking with 1hy and 1.8hy makes sense.

this puts us into the realm of the "large gap" both the .002 and the .004 gap measured pretty close to each other as the frequency changed. IF 22 ohms isn't too much for the bass inductor i'd look to go with say a .003 gap and adjust the turns on #29 to nail 1.8hy at 100hz. Since it is for the lower corner frequency, the extra turns do not bother me so if the DCR isn't an issue it seems like a good start.

for the 1hy. since the corner frequency is 2K i'm a bit wary of the higher turns. Note how the 10K plots go south as wire gauge goes up and gap size goes down. Here the #28 bobbin with the .004 gap looks to be a good choice. The 12 ohms DCR isn't too high but the idea that the 10K plot crosses the others suggests the beginning of a resonance or other "odd" behavior.

neither of the above suggestions take into account the vanishingly low flux levels but that is still uncharted water and i feel we should at least dot the i's and cross the t's before we go beyond.

and yes, the form factor is the same as the autoformers.

dave
dave
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

Actually that all sounds quite good. The 22 ohms is definitely NOT too high... it is actually lower than the Lundahl (25 ohms) and the Sowter (47 ohms). Any resistance below 75 or 100 ohms would work out well. That is one of the degrees of freedom over the LCR, so we can easily take advantage of it.

I agree with your assessment... the real issue is the uncharted water at low flux densities, but I think we'll be OK.

-Steve
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

sbench wrote: Any resistance below 75 or 100 ohms would work out well.
i ended up with about 12 ohms and a .003 gap for the 1hy and 17 ohms and a .0025 gap for the 1.8hy. I agonized over this, but going up in DCR would push the Fres nearer to the point where the filter is intended to work, and going lower in DCR would reveal other nonlinearities that my meter says were there. (i fully understand that my test method may be flawed but you gotta go with somehting)

essentially, i wrote what the meter said on them and they are ready to ship. Hopefully your test method will shed further insight on the situation. Oddly enough, the last two times i did this, the recipients "cheapo" meter measured surprisingly close to what i had agonized over saying. Oddly enough i have two "good LCR meters" and they rarely agree. If the "new" meter works... it should help triangulate the errors.

dave
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

Oddly enough, the last two times i did this, the recipients "cheapo" meter measured surprisingly close to what i had agonized over saying. Oddly enough i have two "good LCR meters" and they rarely agree. If the "new" meter works... it should help triangulate the errors.
My experience more closely matches yours... The most accurate meter for inductance I have is the Owens Bridge critter. But that also shows how much L varies with V and F.

What I'll do is actually place them in circuit and adjust the resistors (starting from "calculated values" until I get flat response). Using Spice, and measuring the response, that method converges on the right solution rapidly.

Will need the 2:1 and 3:1 xfmrs to do that too.

Best Regards,
Steve
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

Inductors and transformers arrived today :D
Will build up one of the PCBs with them.

Steve
Johnny
Posts: 62
Joined: Wed Jan 26, 2005 3:43 pm

Post by Johnny »

Hi Steve,

I can hardly wait for the results with the Slage iron. Please let us in on all the fun stuff.

Johnny
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Johnny wrote:Hi Steve,

I can hardly wait for the results with the Slage iron. Please let us in on all the fun stuff.

Johnny
oops.. it will be a few more days... my idiot packing manager accidentally sent 3-3:1's and a single 2:1. Luchily a new bobbin will show up by the weekend and steve is man enough to make the bobbin switch :-)

dave
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

Ahhh. Here is the Slagle iron version. Parts stuffed except for the LR resistors. I'll try it out this weekend, hopefully.

Steve
Attachments
slalr1.JPG
slalr1.JPG (101.6 KiB) Viewed 17346 times
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

from Dave...
for the 1hy. since the corner frequency is 2K i'm a bit wary of the higher turns. Note how the 10K plots go south as wire gauge goes up and gap size goes down. Here the #28 bobbin with the .004 gap looks to be a good choice. The 12 ohms DCR isn't too high but the idea that the 10K plot crosses the others suggests the beginning of a resonance or other "odd" behavior.

neither of the above suggestions take into account the vanishingly low flux levels but that is still uncharted water and i feel we should at least dot the i's and cross the t's before we go beyond.
Dang Dave. You're good! Plunked in the values for the resistances of the chokes and the transformers into the LR Spice sim, out pops a set of 4 values for the 4 resistors. Plunked those values into the board, and prelim results are GOOD.

I'll be doing a lot more measurements, but initially, with no changes, it looks like the first channel tried comes within a couple tenths of a dB from 20Hz thru about 50kHz. The lower resistance of he transformers WRT other vendors improves the LF cutoff... essentially flat to about 15Hz before heading south.

Steve
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

The lower resistance of he transformers WRT other vendors improves the LF cutoff... essentially flat to about 15Hz before heading south.

Steve
Could you elaborate on this a bit. Personally i thought the transformers might have been a little light on the inductance, but do not see where the DCR of the transformer comes into the picture.

Did you also find the amorphous cores hidden in wiht the packing materials? They will slip on the bobbins and should give you a feel for what that core sounds like if you are so inclined. Do note that the cut on one of the cores is a bit sloppy so the gap might be a bit unpredictable, but it should give you an idea of what the core "sounds" like.

dave
sbench
Posts: 296
Joined: Fri Jan 28, 2005 3:45 pm

Post by sbench »

Could you elaborate on this a bit. Personally i thought the transformers might have been a little light on the inductance, but do not see where the DCR of the transformer comes into the picture.

Did you also find the amorphous cores hidden in wiht the packing materials? They will slip on the bobbins and should give you a feel for what that core sounds like if you are so inclined. Do note that the cut on one of the cores is a bit sloppy so the gap might be a bit unpredictable, but it should give you an idea of what the core "sounds" like.
On the amorphous cores. Yep. Saw them, wasn't sure what to do wid dem, but I'll try that out at some point.

On the DCR... interestingly, it's so simple that everyone forgets the issue..... The DC resistance acts EXACTLY like added plate resistance. We all know that LF response is better with lower plate resistance (f~R/L), but we keep trying to up the L. But you can lower R and achieve the same effect. Well, the lower R improved the LF response so that the corner moves well below 20Hz. Remember I've got a small amount of controlled positive feedback at low frequencies in this design, and it is easy to adjust the Lf response.

Probably be tied up tomorrow; there's a big meeting at work that I'm kinda in the center of, but I'm pretty happy with the initial results. I'm *really* glad the inductor values behave like we thought. (remember the top of this thread?)

Best Regards,
Steve
Post Reply