Here's a replacement for the THAT2252 or THAT4301 RMS detector using a CA3046/LM3046/AS3046 5 transistor array.
This post is from the thread THAT2252 RMS Detector Replacement Using A THAT300 Array
THAT2252 RMS Detector Replacement Using a CA3046/LM3046/AS3046
This is basically a dBx 303-style detector with some modification. The dynamic range is suitable for a compressor/limiter.
A nominal input level of -6 dBu is chosen with the assumption that a 1246 line receiver will be used prior to the detector.
R1 and VR1 provide an approximate +/-3 dB adjustment range for the input "0" reference level.
R2 sets the input reference current at 7.5 uA RMS when the trim is at midpoint.
OA1 and Q3/Q4 are a bipolar log converter. D1 is a bias diode along with with current sink R3.
The scale factor at the outputs of OA1 and the junction of D1 and Q4 are 3.05 mV/dB.
OA2 multiplies positive log results at the output of OA1 by 2X to provide log-domain squaring.
OA3 multiples negative log results at the junction of D1 and Q4 by -2X to invert polarity and provide log-domain squaring.
The output of OA4, when used, level-shifts the outputs of OA2 and OA3 negative by one Vbe.
Q1 and Q2, along with R11 and Ct, are a "log-domain filter."
R11 sets the timing current at -7.5 uA.
OA5 buffers the timing capacitor.
The scale factor at the output, which is the log-domain square of "the mean of the sum of the squares," is 6.1 mV/dB.
Note that the square root operation required to complete RMS conversion - in a compressor application - is performed by the VCA's 6.1 mV/dB scale factor.
The log of the input scales at 3.05 mV/dB but the VCA's exponential scale factor is 6.1 mV/dB.
The difference in scale factors, 3.05/6.10, results in division by two which, in the log domain takes, the final square root.
There are differences between the THAT2252 and 43XX-series detectors.
The THAT2252 has an output that is elevated one Vbe when the input reference current equals the timing current.
The THAT4301 and rest of the 43XX-series use detectors which have a 0V output when the input reference current equals the timing current.
The 4301 "No Vbe Up" detector is better-suited for combined True Power Sum or Quadrature summed detectors because, when they're operating at their nominal level with 0V output, have lower errors due to temperature mis-matching between alternate channel CA3046 arrays.
J1 and J3, for illustration, selects either the 2252 or 4301 mode and can be hard-wired depending on the RMS IC being emulated. When a 2252 is emulated R7 can be removed and the left-hand side of R4 should be grounded.
Pin 13, which is the CA3046 substrate, should be lightly pulled to the negative supply even if the Vbe ref op amp is not used.
The timing cap jumper eases symmetry adjustment with a 'scope and is particularly useful when multiple detectors are summed for True Power or Quadrature Power summing. J2 can be eliminated and linked directly to the timing cap. If a polarized cap is used for Ct it should be tied to -15V by J3 for the "0 Vbe up" configuration to prevent reverse polarization.
The AS3046 arrays are available on eBay or from Cabintech: https://cabintechglobal.com/semi and also from CE Distributing https://www.cedist.com/search/node/as3046
AS3046 Data Sheet: https://cabintechglobal.com/pdf/ALFA_RPAR_AS3046.pdf
THAT2252 RMS Detector Replacement Using a CA3046/LM3046/AS3046
- mediatechnology
- Posts: 5594
- Joined: Sat Aug 11, 2007 2:34 pm
- Location: Oak Cliff, Texas
- Contact:
Re: THAT2252 RMS Detector Replacement Using a CA3046/LM3046/AS3046
Are those arrays still in production or NOS? I used a lot of cheap npn arrays (3086) over the years.
I guess there are some premium matched arrays still around.
JR
I guess there are some premium matched arrays still around.
JR
Cancel the "cancel culture", do not support mob hatred.
- mediatechnology
- Posts: 5594
- Joined: Sat Aug 11, 2007 2:34 pm
- Location: Oak Cliff, Texas
- Contact:
Re: THAT2252 RMS Detector Replacement Using a CA3046/LM3046/AS3046
The AS3046 is in current production and stocked. It has tighter Vos and Ios limits on the differential pair and less delta Vbe on the uncommitted transistors than the original and runs about $4 US at qty 4 units.
There is a shrunk feature GHz ft LM3046, the Renesas HFA3046, available in smt for RF. https://www.renesas.com/us/en/document/ ... -datasheet
There is a shrunk feature GHz ft LM3046, the Renesas HFA3046, available in smt for RF. https://www.renesas.com/us/en/document/ ... -datasheet
Re: THAT2252 RMS Detector Replacement Using a CA3046/LM3046/AS3046
Kool... that looks like the better 3086.
I noticed in my TS-1 that the Rbb of the small geometry devices in the 3086 caused a major fraction of a dB error in my dB meter at +20 dBu (due to resistance in series with base-emitter junction).
I probably could have compensated for that in the design, but there were other things higher up on my list to improve.
There are probably other modern arrays with lower Rbb (its only money ).
JR
I noticed in my TS-1 that the Rbb of the small geometry devices in the 3086 caused a major fraction of a dB error in my dB meter at +20 dBu (due to resistance in series with base-emitter junction).
I probably could have compensated for that in the design, but there were other things higher up on my list to improve.
There are probably other modern arrays with lower Rbb (its only money ).
JR
Cancel the "cancel culture", do not support mob hatred.
- mediatechnology
- Posts: 5594
- Joined: Sat Aug 11, 2007 2:34 pm
- Location: Oak Cliff, Texas
- Contact:
Re: THAT2252 RMS Detector Replacement Using a CA3046/LM3046/AS3046
Turns out CE Dsitributing also carry the AS3046 and full Alfa Rpar lineup.
https://www.cedist.com/search/node/as3046
Alfa Rpar also make the AS194/394 (LM394) super-match transistors and and the CA3080/3280/LM13700/13704 OTAs.
For almost any THAT2252 functional replacement in mass production the 43XX-series analog engines in SMT make the most sense.
For DIY and boutique small run rolling one out of a DIP 3046 may make sense for many folks.
In that situation by the time a 43XX is placed on a DIP adapter the expense weighs in favor of full through-hole.
The 43XX mounted on a DIP adapter itself is not super-expensive but what I found out was that the pin strips needed to mate with a conventional socket are not. Sqaure post headers are inexpensive and often used but you can't mate them into a DIP socket.
John you should visit the Cabintech semi page. You'll be amazed at what is still being made including MN-series BBDs, Vactrols etc: https://cabintechglobal.com/semi
https://www.cedist.com/search/node/as3046
Alfa Rpar also make the AS194/394 (LM394) super-match transistors and and the CA3080/3280/LM13700/13704 OTAs.
For almost any THAT2252 functional replacement in mass production the 43XX-series analog engines in SMT make the most sense.
For DIY and boutique small run rolling one out of a DIP 3046 may make sense for many folks.
In that situation by the time a 43XX is placed on a DIP adapter the expense weighs in favor of full through-hole.
The 43XX mounted on a DIP adapter itself is not super-expensive but what I found out was that the pin strips needed to mate with a conventional socket are not. Sqaure post headers are inexpensive and often used but you can't mate them into a DIP socket.
John you should visit the Cabintech semi page. You'll be amazed at what is still being made including MN-series BBDs, Vactrols etc: https://cabintechglobal.com/semi
Re: THAT2252 RMS Detector Replacement Using a CA3046/LM3046/AS3046
or not.... these days my specialty is home brewed beer..mediatechnology wrote: ↑Thu Jun 13, 2024 1:08 pm
John you should visit the Cabintech semi page. You'll be amazed at what is still being made including MN-series BBDs, Vactrols etc: https://cabintechglobal.com/semi
JR
Cancel the "cancel culture", do not support mob hatred.