Take the audio right at the grid pin of the 12AT7 (V13) socket. I use a 9-pin plug that replaces the tube. The 6AQ5 is pulled out to save drain on the power supply and reduce heat. Originally, I just ran a shielded wire from pin 2 (grid) and pins 4,5 (gnd) to the outboard amp, an old 50's vintage 10 watt mono hi-fi amplifier with a pair of 6V6's in the output. This allows use of the internal receiver AF gain control. Highs and lows were still missing, although the distortion was reduced through the outboard output stage. Just run the shielded lead out through one of the ventilation louvres in the receiver cabinet.
High Frequency Improvements
Further investigation showed that the high frequency response was being killed by C147 and C148, 510 pF micas that bypass the highs as they pass through the noise limiter stage. These were add-ons by Collins, missing in my older receiver (s/n 1500 something) but present in my later one (s/n 2800 something). As further evidence, the caps are numbered out of sequence with other nearby caps on the schematic. I simply clipped out the caps, and the result was flat response out to at least 7 kHz. The 6AQ5 stage has a 0.01 uF cap (C103) right across the output transformer. If you want to improve the high end of this stage, try replacing C103 with a 0.001 uF. On other receivers with a similar cap, including BC sets, complete removal often results in high frequency oscillation. This is especially a risk with the feedback circuit in the A4. All the more reason to abandon the 6AQ5 stage to outboard audio.
Low Frequency Improvements
The low-end response was still missing. The schematic shows that the noise limiter diodes are in the circuit at all times, and when the limiter is "off" the diodes are simply biased on with enough voltage that even the strongest audio peaks will not turn the diodes off to cause clipping. The problem with the low frequency response is the noise limiter coupling caps C93 and C96. The original value of each is 0.01 uF. I replaced these with 0.1 uF caps. Actually, I just bridged the new caps in parallel across the existing caps. Using this technique is much easier, since Collins just about ties a square knot with the wire leads around the solder lug terminals before applying solder. And, if the original caps aren't leaky they will have no effect on the circuit if they just stay in place, resulting in less heat damage to nearby wires and components. Of course, you could just clip them out, leaving the wire ends in place. This mod brought the low end response flat to about 30 Hz. If you plan to use the stock 6AQ5 audio amp, you will need to increase C100 and C101 to 0.1 uF. One caution: If the receiver is an early s/n, it may be necessary to install the Collins modification designed to reduce hum in the audio, once the frequency response below 200 Hz is improved. This mod is described in the compendium of 75A4 modifications, available from Electric Radio.
Cathode Follower Output
I further modified my arrangement over the holidays this year. I suspected the shielded lead, about 3' long, running from the 12AT7 socket to the amp, had enough capacitance to shunt some of the highs to the amplifier. The impedance at this point is 500k, as determined by the AF gain pot. I built a simple cathode follower using a 6AT6 triode. It mounts in a small metal box, with a 9-pin plug on the bottom and a 7-pin tube socket on the top, and all components fit inside the box. The B+ is taken from pin 5 of the vacant 6AQ5 socket using a 7-pin plug, letting the original AF output transformer serve as an additional filter choke, adding 10 uF of filtering (in the adaptor box) with a miniature 10 uF/350v electrolytic filter cap I found in the junk box. A 10 uF non-polarized electrolytic is used as a blocking cap at the cathode follower output, although this is probably overkill. Unless you plan to attach a low impedance line to the output of the follower, a 0.1 uF will effectively couple to the high impedance input of the hi-fi amp. Details on the 6AT6 cathode follower can be found in the RCA receiving tube manual. I used an RCA "phono" plug on the adaptor box to attach the shielded line. The cathode follower seems to result in slightly better response, both high and low end, at the expense of about 3 dB of audio gain.
The down side of using an outboard amp is the likelihood of multiple ground loops. With the cathode follower, I ended up using an isolation transformer between the receiver and the amp. I stuck in a UTC "O" series interstage transformer, 15K to 60K, with the 60k winding going to the amp. This compensated for the gain loss at the cathode follower and brought the hum level down. The ground loops are caused by the necessity to shield the audio lead from the receiver to the amp, while attaching grounded coax to the antenna terminal, and the outboard amp is rack mounted, thus having its own ground to the station infrastructure. I have the hum level so low that it is noticeable only with headphones when the AF gain is turned all the way down. This is much lower than the amount I initially experienced with the stock receiver, despite their deliberately killing the low frequency response below 200 Hz with the 0.01 uF caps. I suspect hum in the audio was one of the reasons for the under sized coupling caps, along with their affinity for "communications quality" slopbucket audio.
I also have another outboard box, that plugs into the "A" mechanical filter position, allowing the selection of additional selectivity filters. At present I have 300 Hz, 3.1, 4, 6, 8 and 16 kHz filters available. But that is a whole other story.
I hope this helps anyone contemplating improving the 75A4 audio, which was apparently little more than an afterthought at Collins. The above mods give pretty decent AM audio with the A4, although it is still not perfect. Not much you can do about the inherent distortion caused by the mechanical filters, as anyone who has compared an original R-390 with an R-390A will attest.