|
|
Post by svart on Jul 10, 2018 14:40:00 GMT -6
10 bucks for the board? I'm in. Haven't built anything in a while, could use an LA3A style comp.
Edit: I'd go for 2 if you have enough.
We'll see! I can always have more made I suppose. I have to buy them in lots of 10. Sounds like this lot might already be spoken for. I think the hardest part to get will be the EL panel. I bought mine from Audio-Scape via Ebay. They still have them listed on their website for 13$ each. I don't have any affiliation with them, they just came up in a search and the price was right. www.audio-scape.com/products/el-panelThe NSL-5910 CDS cells (the ones all the replacement guys are using) are also a bit harder to get as only one place seems to have them currently for about 5$ each: www.alliedelec.com/luna-optoelectronics-nsl-5910/70136770/ |
|
|
|
Post by winetree on Jul 10, 2018 18:31:45 GMT -6
Interesting that you borrowed the T4B replacement design from Peter Purpose's LaZ-La3A from GroupDIY. His PCB has been discontinued for years, but the idea for dealing with the optical cells is cool. Nice Lil project! Are you gonna do a front panel for it?   |
|
|
|
Post by NoFilterChuck on Jul 10, 2018 19:11:20 GMT -6
I built a pair myself!  |
|
|
|
Post by svart on Jul 30, 2018 11:15:15 GMT -6
 Took an hour over the weekend to get some of my chassis built up to test the PCBs. I hope to have some audio through one of them in the next day or so. |
|
|
|
Post by svart on Jul 30, 2018 19:56:28 GMT -6
Audio!
Blinky lights!
So it seems to work. I get audio through, and I get blinking EL panel and a little compression.
Even with things turned all the way up I don't get a ton of compression, but I haven't tried adjusting any trim pots yet, they're all centered so far.
The one thing that is strange is that I'm using a 15VA transformer and it's getting warm.. It shouldn't be getting warm. I guess I need to check around and see if anything is amiss.
|
|
|
|
Post by matt@IAA on Jul 30, 2018 21:44:35 GMT -6
Psst svart in that photo I think you’ve forgotten to solder the components onto the pcb. Just tryina help.  |
|
|
|
Post by svart on Jul 31, 2018 6:08:53 GMT -6
Psst svart in that photo I think you’ve forgotten to solder the components onto the pcb. Just tryina help. Everyone's a critic. I used blanks to find my arrangement. |
|
|
|
Post by svart on Jul 31, 2018 18:47:57 GMT -6
So yeah, Everything is in line now. A wiring error on my part was the source of the warmth..
Anyway, I'm not getting more than about 50V to the EL panel at max reduction and max stereo adjustment. It's roughly 7V into the autotransformer and then about 49 out with a 1K sine at about -6dB input. Seems like the autotransformer isn't stepping up as much as I'd hoped.
I suppose I'll get a pair of Hairball's LA3A autoformers and give them a try. Otherwise I try to find any information on output voltage from the sidechain to see if my circuit is acting normal, but I think it is.
I tried a few different transistors in the sidechain and it seems like it doesn't really matter what's used there. I designed for 2N5089, but also tried 2N3417 and MPSA06 with almost no difference in results.
If it comes down to it, I could change the 68K feedback resistor to get more gain from the sidechain, but I don't think I should need to.
|
|
|
|
Post by svart on Aug 1, 2018 13:32:34 GMT -6
BTW if I can verify that it's my autoformer choice that's causing the seemingly low EL voltage, then I'll whip up a one-pager document for the PCB and start selling them for those interested.
It's just a straight schematic copy, but there are a few tidbits that can be addressed.
One is that I used modern CBE pin arrangement on all the TO-92 parts like 2N5089. The 2N3417 use the old style that swaps the (B)ase and (C)ollector. You can use the original parts, but with twisting the B and C legs. Alternately I found that the cheaper MSA06 works just as good as the 2N3417.
There are a few resistors you can change for less gain that are the same as the famous "gain switch mod". There is a place for switching the sidechain output section to class AB from class B by using diodes and a different resistor value.
|
|
|
|
Post by svart on Aug 3, 2018 18:34:59 GMT -6
Good news!
I got my autotransformers from hairball, and they have a bit more output.
I do think the speaker matching transformers will work pretty well if you don't need more than like 8db compression.
The autoformer clones from hairball put out 10x the ac voltage. The quam tblu transformers seem to put out around 8x for the same signal.
I know that someone had edcor wind an autotransformer for this, so I inquired about it, and the costs.
It might be possible to get a total edcor transformer solution for a lot less than anything else.
And yes, edcor makes quality stuff, and likely better than the originals for sure.
That being said, I'm going to open up sales of the boards next week. I've started a document full of information and pictures, but please don't expect total build instructions like some other kits.
If edcor comes back with good prices for the autotransformer, I might buy a handful of them to sell with the boards since a bulk discount can be passed on.
Right now I'm guessing a channel will cost around 100$ to build with the transformers and shipping all included, but without the case and power transformers.
A cheap BUD aluminum case is around 50, a power toroid is around 17$.
I'd say less than 300$ for a stereo set in a single chassis if you do all the work yourself, get some cheap used VU from eBay, and print your own front panel stickers.
Could be a lot cheaper if you have parts already.
PCBs will be 12$ each plus shipping.
|
|
|
|
Post by bram on Aug 4, 2018 14:33:55 GMT -6
Interested in 1 or 2 boards depending on availability.
|
|
|
|
Post by svart on Aug 4, 2018 14:50:55 GMT -6
Interested in 1 or 2 boards depending on availability. I'll have 18 boards, and can always get more. |
|
|
|
Post by jcoutu1 on Aug 4, 2018 22:05:45 GMT -6
Interested in 1 or 2 boards depending on availability. I'll have 18 boards, and can always get more. How much for a built stereo unit? |
|
|
|
Post by svart on Aug 5, 2018 7:55:48 GMT -6
I'll have 18 boards, and can always get more. How much for a built stereo unit? All you got. |
|
|
|
Post by wiz on Aug 5, 2018 15:26:32 GMT -6
How much for a built stereo unit? All you got. In, I ain’t got much 8) Seriously what about everything but the case? cheers Wiz |
|
|
|
Post by svart on Aug 6, 2018 9:05:54 GMT -6
In, I ain’t got much 8) Seriously what about everything but the case? cheers Wiz Let me see if edcor comes back with a decent quote for the autotransformer. If they do I'll spec an output transformer and see if they do a package. If they do, it's a lot more likely I could do a partial kit or something.. So I hooked up a random VU meter I had from a set and messed around for a while before realizing that it was broken.. and then grabbed another to find that it doesn't quite zero with the circuit. It's not unheard of especially since I don't even know the specs of these VU meters, so once I calibrate a board I'll start selling them. |
|
|
|
Post by svart on Aug 6, 2018 18:46:23 GMT -6
More updates! I only had a few minutes to tinker tonight but I worked on the VU.
I changed a resistor and got my meter to zero pretty easily. Now the GR scale is off. I get about half the GR from the meter than I show coming back in the daw.
So let's say I put in -10db signal, and turn the compression knob to get about -15db out, I read 2.5db of GR. So for an output of -20db, I see 5db, etc.
Apparently I can get almost 20db of GR from the unit. Quite impressive!
I think this is all normal for using an unknown Vu meter. If I can change the VU adjust pot to something higher, I can probably get it right on the money.
I'll add the resistor changes to my document in case anyone cares. If you buy a kit from someone like hairball, the VU meters shouldn't need this kind of adjustment.
|
|
|
|
Post by matt@IAA on Aug 6, 2018 19:03:29 GMT -6
Depending on where the meter signal is coming from could the difference be due to log behavior vs linear?
|
|
|
|
Post by svart on Aug 6, 2018 19:24:30 GMT -6
Depending on where the meter signal is coming from could the difference be due to log behavior vs linear? I'd say yes it's possible, but it's more likely that they're just different current than what was in there. I say that because the meter-zero sets some kind of baseline DC, and the meter-adjust seems to set operating current. I essentially had to half the feeding resistor, and then double the adjustment resistance. Now it tracks 1:1 with GR. I say these things because a cds cell like the ones used here don't quite react in a linear fashion, but not quite log either. I'd say it's more log than linear though. I also assumed that most VU meters are log, since a lot of them are used for direct audio level. Since VU meters are all current driven, I also could have used a known voltage source with a pot to figure out what resistance and voltage gave me full scale movement, and I could figure the current needed from that using ohm's law. Once I had that it's easy to measure the voltage at minimum and maximum scale in the unit in conjunction with the needed current and figure out the min and max resistors need to get me in the ballpark.. Or I can do what I did and take an educated guess..  I did go ahead and change the adjustment trim pot out for something higher, and found it aligned much better with 1k more adjustment, so my problem is solved here. |
|
|
|
Post by svart on Aug 6, 2018 21:06:47 GMT -6
I guess I should also clarify that the meter's motor itself is a *linear* device, but the printed scale is *log*.. The CDS cell is being used as a low-side variable resistor, so it's pulling down the voltage through the meter-zero circuit forming a voltage divider that sets the static voltage/current. The meter-adjust trimmer replaces a discrete resistor in the original circuit in order to have a range to set the operating current of unknown meters.
Light power, measured in Lux, is also log. I'm not sure I've ever seen the EL panel voltage curves, so I don't know how they respond at all!
|
|
|
|
Post by matt@IAA on Aug 6, 2018 22:08:02 GMT -6
Right. I was just thinking in a VCA the sidechain voltage is linear to dB reduction. Not sure where the metering signal is coming from, but my thought was there may be a log measurement representing a linear result or vice versa. Anyway glad you got it sorted.
|
|
|
|
Post by svart on Aug 7, 2018 7:30:15 GMT -6
Right. I was just thinking in a VCA the sidechain voltage is linear to dB reduction. Not sure where the metering signal is coming from, but my thought was there may be a log measurement representing a linear result or vice versa. Anyway glad you got it sorted. In these, the sidechain is mostly the same as the audio path in that it's simply a power amplifier, but instead of increasing gain to an output, it increases gain(AC voltage) to the EL panel. The "gain reduction" knob is roughly the same as the "gain" knob in the aspect that higher gain through the sidechain makes the panel becomes brighter with increased signal, and thus makes the CDS devices (both configured as attenuators) increasingly conduct and pull the audio and meter feed signals down. It's a glorified optical feedback with the audio AC being used as a CV and using the EL panel as a makeshift lowpass (averaging) filter. I've seen some T4B come with a series capacitor to limit low frequency content, likely to spectral-ly tailor the control voltage so that the EL panel response is similar over frequency. I'd bet that if I did EQ sweeps, I'd find that there is a tilt effect where the EL panels are more sensitive at lower frequencies. I did snoop around online to see if I can find brightness curves for the EL panels, but I don't see any direct measurements for something similar to the T4 panels. Some anecdotal graphs for generic panels seem to show anything from purely linear to antilog curves depending on panel materials and design. However, I did find some of Kenetek's graphs that show a roughly linear output from the T4B over a linear input voltage. This makes sense, since the meters are linear devices but with log scales. I'm sure that they used whatever they had available at the time (doubtful they spec'd special parts), and there was little consideration for the actual T4 panel response, other than to ensure the audio output was log and repeatable. |
|
|
|
Post by svart on Aug 9, 2018 11:20:40 GMT -6
So I think I'm ready to sell these things. I have a BOM and most of the erratia and information sheet done. I don't have build instructions, but it's a simple build and anyone who's done any clones will probably be OK.
|
|
|
|
Post by svart on Aug 9, 2018 11:30:55 GMT -6
So right now I have these members and quantities from this thread and PM's: winetree x2 jeromemason x1 matt@IAA x2 johneppstein x2 bram x2 wlinart x2 That's 11 boards of 18 spoken for unless I have overlooked anyone. If I have, please get with me ASAP! I can always get more boards in about a week's time if more people want them. I'll finish the files and make a download available in the next few days and we can move to payment/shipping. |
|
|
|
Post by bram on Aug 9, 2018 14:06:51 GMT -6
Great, thanks Svart!
|
|
