DRIP Fairchild 670 Lat/vert issue?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

remsouille

Well-known member
Joined
Oct 15, 2012
Messages
218
Hi all,
I've been using my V3 670 for quite some time now and it's sounding marvelous (no shit, sherlock). However, I had not really used it on Lat/vert mode since today, and as I put it across my mix buss just a couple of hours ago, I noticed something was wrong right away...
I ran sweeps through it, and  it turns out there is quite a big HF roll off on Lat/vert mode, about -3db at 20K, starting around 1K. Dual mono is perfectly fine, the problem is on both channels on lat/vert. The unit is also around 3db quieter on lat/vert, which may be normal, I don't know. I know that the lat/vert switch only switches the ins and outs to a standard passive M/S BBC-style matrix, but I really don't see how that would affect the frequency response so bad...
Any ideas?
 
While I'm absolutely not an expert on tube compressors, if the difference between L/R and M/S (or Lat/Vert) is a passive transformer matrix, could the change in frequency response be related to the load on the transformer secondaries? Maybe fiddling with termination resistance or zobel networks would help even things out?

No idea about the gain issue, though it may also be related to loading on the transformers. I'd understood that M/S matrices normally add gain though, simply based on the maths involved...
 
TwentyTrees said:
While I'm absolutely not an expert on tube compressors, if the difference between L/R and M/S (or Lat/Vert) is a passive transformer matrix, could the change in frequency response be related to the load on the transformer secondaries? Maybe fiddling with termination resistance or zobel networks would help even things out?

No idea about the gain issue, though it may also be related to loading on the transformers. I'd understood that M/S matrices normally add gain though, simply based on the maths involved...

Or could it be that somehow there is a slight phasing issue between the two channels? The bass wouldn't be affected too much but it would be more audible as the waves tighten up? From what I gathered from Greg at drip and by watching a couple of videos, it seems normal that the 670 has less gain in M/S mod, though I don't really understand why..
 
I would vote for transformer interactions.  Differences in the transformers themselves, wiring within the unit, layout, etc. 

L/V mode, the center image information is greater in level than either dual mono or side, and thus causes either more compression or requires a lower input level to avoid over compression, thus leading to a lower output level once recombined to L/R. 
 
emrr said:
I would vote for transformer interactions.  Differences in the transformers themselves, wiring within the unit, layout, etc. 

L/V mode, the center image information is greater in level than either dual mono or side, and thus causes either more compression or requires a lower input level to avoid over compression, thus leading to a lower output level once recombined to L/R. 
This is the schematic, in case you're not familiar with it http://www.4tubes.com/SCHEMATICS/Pro-Audio/Compressors/Fairchild670.JPG

The drip V3 fairchild is very well laid out, with both channels identically laid out on the main board and I'm using the expensive sowter transformers, so I doubt this is a wiring/transformer/layout issue. I must have done something wrong, I just have yet to find what...

Doug, looking at the schematic, does that loss in gain make sense to you? I'm not sure exactly what you're implying with your post.
Thanks for chiming in!
 
Can you say for certain that it is very well laid out for high frequency interactions?  I haven't seen it, and can't say.    At any rate, it doesn't take much of an error to cause high frequency interactions which would roll off treble.  There is no reason to say either that Sowter transformers would be immune from these problems.  They are not necessarily designed for L/V interweaving, unless I have missed a specific statement in their data sheets.  When Mr Narma refers in interviews to the criteria with which transformers were chosen, he mentions winding balance.  Any bits of wiring layout are extra wild-cards that can possibly change response. 

Is the gain difference found without compression active?  Same input level? 
 
emrr said:
Can you say for certain that it is very well laid out for high frequency interactions?  I haven't seen it, and can't say.    At any rate, it doesn't take much of an error to cause high frequency interactions which would roll off treble.  There is no reason to say either that Sowter transformers would be immune from these problems.  They are not necessarily designed for L/V interweaving, unless I have missed a specific statement in their data sheets.  When Mr Narma refers in interviews to the criteria with which transformers were chosen, he mentions winding balance.  Any bits of wiring layout are extra wild-cards that can possibly change response. 

Is the gain difference found without compression active?  Same input level? 

What I can say, is that the transformer set I goty from sowter is specially designed for drip's 670, so I assume they should be up to spec.
Yes the gain difference is at unity, without compression!
Thanks!
 
Hi all, there has not been much progress on that issue. It seems that the only difference in signal path between the Lat/vert and left/right mode is that the signal goes through the relays on the agc card. Is it possible for relays to have an impact on frequency response?
 
No. It's most probably about different leakage inductances across the different windings of the transformers when set up in matrix mode. Unless you use transformers that are fully symmetric and wound on a single bobbin, you run into poor coupling..

Jakob E.
 
Were they made using the original winding geometry? (Did CJ ever hack these?  :) )

Were they ever actually tested by Drip when developing the kit?

Jakob E.
 
Being the naïve fellow I am, I assume he did. But maybe I'm wrong. I did ask him about the issue but I haven't heard from him on a while...
 
Hi all, I enquired Sowter about my issue and this is what he had to say:

"Both the input transformer (type 8346) and the output (type 8344) are geometrically balanced.  We did this because it is important that everything is resistance and capacity balanced so the matrixing works properly. (my opinion)

The output transformer uses a bifilar would secondary to achieve the balance.  The input transformer uses a two chamber bobbin with two secondary and one primary section in each chamber.  The 4 secondary sections are connected in series.

For your information we had one customer raise similar concerns to yourself some years ago and we made an alternative  option to our standard design.  We added an additional primary winding to each side bifilar wound with the first.  The primary windings are connected in series across the bobbin resulting in two primary windings with the same turns as before but improved coupling with all four secondary sections.  This is of course a trade of because of the capacitive coupling between the primary windings.  The customer has purchased  several pieces of both types but we do not have any good data comparing the two approaches.

On the understanding that you would be prepared to share your results we will be happy to exchange your current type with the alternative design."

They did send two replacement input trafos to me, I installed them today, and the issue seems to be gone! I'll run some further testing in the next few days to make sure everything's ok, but so far, so good!
R.
 
Sowter are real guys. Respect.

Make sure to share all aspects of your build with them - this way you'll save many other builders a lot of work and worry..

Jakob E.
 

Latest posts

Back
Top