ACME 4-tube Reflex
Broadcast Receiver (1924)

Curiosity got the best of me when I first saw this circuit.  I've only built one Reflex receiver in the past and it was a relatively simple 1-tube set using a 3A8GT.  I like the idea of reflexing to make one tube do multiple functions but the old timers will tell you straight up that the technique has its share of  problems and quirks.  Well, thats all I needed to hear as MORE incentive to take on this challenging project.

This fella works as follows:
V1.  RF amp Stage #1
V2.  RF amp Stage #2, later reflexed as an audio amp stage
V3.  RF amp Stage #3, later reflexed as an audio amp stage
Crystal Detector, then back to V2
V2.  Audio amp
V3.  Audio amp
V4.  Audio amp
to speaker or phones

See schematic for more detail.

I started parts shopping with the Acme 4-tuber in mind.  All I had was an ACME tuning cap in the junkbox.  I drifted away to some other projects and lo-and-behold a 4-tube homebrew version popped up on ebay at a very good price. ($36, only 2 bidders - that speaks for the condition it was in - that breadboard below fetched ~ $500 !!!).  It was indeed a "parts set" but provided most of what I needed for my project.  No, I didn't butcher up a perfectly good piece of vintage equipment - somebody beat me to that by chopping out parts and attempting to convert it to a simple TRF set - but I can't call it a restoration OR a homebrew.  Its somewhere in between.  How about restoration of somebody else's homebrew that had been modified?


Here's the one I got.  Not bad but not good either.  Broken sockets, missing parts, faded front panel.  Parts set.

Here's a nice breadboard version of the set that recently sold on ebay (not mine).

Before finding my set I had embarked on making my own replicas of the ACME RF Transformers.  I understood that Acme staggered the center frequency for the R-2, R-3 and R-4.  This is partly due to their inability in those days to make a good flat RF transformer.  I found with a piece of ferrite rod I could wind a pretty doggone flat one but as I learned afterwards that may well not have worked in this circuit.  Part of Acme's insight was to stagger the rf gain distribution as well and my "better" xfmr may not have allowed this.  I could probably have flattened them more with added resistance across the windings and in the process knocked off some of the gain.  That should work but once I had the actual parts there's was no reason to re-invent that wheel.

Just to give an idea of how staggered the R-2, 3 and 4 are, here's a chart of the measured inductances.  Not only are the frequencies staggered but the ratios seem to be as well.

R-2 Pri
300 uH
550 uH
3000 uH
4200 uH
830 uH
2150 uH

My set had an open R-2.  No biggie, these Acme xfmrs are fairly easy to repair.  I previously had bought a 'dud' R-2 from PTOP to learn about the innards so I was ready for this.

My set also had two replacement AMRAD xfmrs and doggone it one of them was open.  I restuffed it with a cheapo 2.5:1 modern xfmr.  One of the quandries I faced regarding the audio transformers was which one could be expected to be the best load for my crystal detector since each of the three were different.  I opted to use the single remaining original Acme one figuring the result couldn't be any worse than original.  If I were to do this from scratch I think I'd go with a high ratio here (T1), something like a 6:1.  I used my cheapo restuffed one for coupling the final audio stage.

So here's the project close to completion:

I guess that faded black panel doesn't look TOO bad in brown.

I completely disassembled the set for cleaning and rewired it. 

Here's a link to my schematic.

The set took right off and worked the first time I fired it up.  I used a modern glass diode for the detector while waiting on parts to build a 'rock' detector.  It was immediately more than I expected out of the set.  A few issues though:

1)  Overload by my local station on 1370.  I finally built an inline trap which I have needed for my projects and that worked wonders.  My Acme is using a tapped variocoupler instead of the recommended loop antenna so I am somewhat selectivity challenged. After all, its only a single-tuned circuit. Once its all tweaked out and at the right regeneration point its plenty selective on the bottom 3/4 of the band.

2)  Tuning range.  The original builder used a generic variocoupler and its not the best combination for the 500pf Acme tuning cap.  Range was about 480-1400 kc.  I removed a couple of turns but it didn't move appreciably.  Removing more would involve some butchering of the coil so I left it be.

3)  Dual pot/rheo.  The Acme info that I have doesn't spell out exactly which of their dual pot/rheo models is to be used in this circuit.  The one found in place had a 30 ohm rheostat which suggests the original guy used 99s (although 01As were in the sockets).  Maybe he didn't know better.  I got one from PTOP with a 6 ohm rheostat.  But...the 30-ohmmer had 200 ohms for the pot section, the 6-ohmmer has 100 ohms.  Again, not specified by Acme but it would seem higher would be better since its a load directly across the A terminals to pick off bias.  I'm not running off of a battery so I guess its moot.  Some of the Acmes use a 2k pot across one of the rf xfmrs as a 'volume' control.  Thats not the case here. The little Acme layout diagram isn't very clear about the connections and I had no schematic.  The pot/rheo has some internal connections which must be noted.

First tests were running the set with about 100v B+.  I tried the 3v C battery both at the end of the pot and on the wiper with differing effects.  First point of confusion was the wiring of the pot itself but after I sorted that out I was able to drop the B+ current draw from the hungry 30-35ma range down to 22ma which seems more appropriate for a 4-tube set.

The bigger surprise came when I lowered the B+ to 70 volts. I noticed when I switched off the radio that I got a burst of volume.  So I dropped the B+ down to about 70 volts and got better performance and a current of only 11ma!

Then I took out the 3v battery and tried 45v drawing 16ma and still sounds good, better than at 100 v.

Filament control behaves a little smoother as well.  It drops dead below 3.5 volts but I seem to get my best/cleanest volume right around 3.8v.  Before it was taking the max on the rheostat.

One suggestion made to me was to set it up for 90v on the last tube which is audio only and maybe run everything else at 45.  That might be the best of both worlds.  I think it would also be worthwhile to add a filament rheostat for the audio stage alone since the other one is somewhat consequential in the regeneration and performance of the radio.  I don't think I can make those changes now in my set but if the reader is building from scratch he'd be well advised to consider that option.  For mine I have decided that 4.5 volts of C-voltage and something around 80 volts B+ seems to a good compromise.

The crystal detector was missing from my set so that was something else I had to improvise.  It had originally used a Brownlie panel mounted type and I'm not able to find a replacement.  So for the meantime I cobbled together the one shown made with some parts sold as an ebay item.  That item had so inherent shortcomings so I added a direct jumper from the shaft of the ticker to the connection point using a flexible coil of phosphor-bronze dial cord.  I also added a strand from that same dial cord to serve as the actual tickler instead of depending on copper wire for that purpose.  It works quite well and is not at all tedious to find an active spot on the crystal.

 Bill Meacham - August 2006