Surplus store score! Collins 26c and 2x Gilfillan Bros/Bendix filters pics

emrr · Aug 29, 2012

I'll have to take some more measurements of mine for comparison.

plumsolly · Sep 8, 2012

I decided to have a go at fixing my original Collins input tx. I figured it was already busted, so I didn't have much to lose. I froze it and dug out the potting. The problem was just one intermittent secondary, which could be affected by tugging on the wire, so I though it was probably an issue right where the lead met the coil. The following pictures are not for the faint of heart - Doug are you sitting down?

I did find an open connection at the point in question. I dug into it a little to see if I could find the coil to reconnect it. I didn't have any luck and may have damaged it further messing around in there, but I am not any worse off than when I started - I could still send it to Doc Hoyer to rewind or to CJ to dissect, if I ever get around to it.

In the mean time, I have really been enjoying my fuller-bandwidth 26c and I think I am going to give the Lundahl a try.

Best,

Ben

emrr · Sep 8, 2012

take some pics of the case itself too.

plumsolly · Sep 11, 2012

Case pics. Not too exciting, but it seems to be made of cast iron, which is neat.

Update on the plots of different input transformers: The Edcor was indeed broken - open secondary winding.

I went ahead and ordered a Lundahl LL7905. I will be back soon with a report on that.

Best,

Ben

plumsolly · Oct 9, 2012

Startling discoveries! A shocking admission! And one dramatic conclusion (sort of)!

I received my Lundahl LL7905 and started to do some testing with it. The first test I ran was with it wired as 1:11 with a 10K/600 pad. 1:11 is pretty close to the original, which is around 1:11.5/500:80K. The test showed some high frequency roll-off that I wasn't expecting (see below)

I decided to run the same test without the pad to see how it affected the response:

Notice the vastly improved response. At this point I went and investigated the pad, which is located on a mic/line switch in a breakout panel I have attached to my 26C. The panel also houses a bypass as well as attack and release controls. Upon inspection, I realized I had wired the pad incorrectly! I had managed to put the 600k resistor on the wrong side of the 5.1k resistors, making an upside down U-pad. With this arrangement, the input of the limiter was seeing a 10.6k output impedance, and my sound card was seeing a 600 ohm input Z - pretty far from ideal. I fixed it and resumed testing. Here is the same test with Lundahl and the correct pad:

This last one is very similar to the no-pad response - much better. But it is still not as nice as what I was seeing with my 600:600 UTC HA-108X with no pad and a little frequency compensation, seen here:

So what does my 600:80K UTC HA-101X look like with the correct pad? Here it is with the secondaries in series and no frequency compensation:

Much better than with the incorrect pad, but still shows the dip caused by having the secondaries in series. So how about the HA-101X with the secondaries in parallel? Here it is with some frequency compensation placed after V1:

Much better! But remember all these test are without the limiting engaged, so lets see what happens when the limiting circuit is in:

CollinsHa101xparallelCORRECTpadlimitinginnofreqcomp.png

I guess I won't be needing that high-end rolloff I had introduced. I am at a pretty good place now - the frequency response exceeds the specs in the manual, probably due to the high quality input transformer, but I have lost some level on the front end. I went from a 1:11.5 transformer with around 21dB gain to a 1:5.8 transformer with 15dB gain. Will I still have enough level with my 10k/600 (25db) pad? I decided, at least for now, that I would - I had lowered the threshold and raised the side chain signal by fooling around with V3, and I can still get it into limiting just fine.

It is back in the rack now and sounding really good. Thanks to everyone who stopped by and helped out.

Best,

Ben