Sunk Cost Fallacy (re: Leak Preamp)

BuzzK

BuzzK

AK Member
Subscriber
The Challenge

I was received a Leak Point One Stereo Pre-Amp and a separate FM Mono Tuner from a generous soul. IMO the pre is cosmetically elegant, especially for its era (Sept. 1959). I thought I'd try to get it going and give it a listen.

The power of for the pre is derived from a Leak power amplifier, which I no longer own. So I built a power supply within a computer PS enclosure to to provide about 185 VDC B+ and 12V for the filaments (which required rewiring), using an umbilical cord to connect the two. Next I replaced the capacitors and most carbon comp resistors in the pre. I replaced several of the short input jacks with modern ones, and repurposed the RECORD OUT to LINE OUT.

The circuit is unusual . . . four EF86s, two per channel. All signals (phono & line level) go through the first EF86. RIAA and selective attenuation is done by negative feedback. The second EF86 serves the tone controls.

There are some inherent problems with the circuit design.

• The EF86s are high gain and high impedance, so they can't drive long cables or a low impedance amplifier.
• The output signals ran through the original umbilical cord to the power amp
• The gain of the unit was designed to be low because Leak used lots of NFB and had high sensitivity in their amps
• The RIAA compliance is poor -- off by several dB according to one source
• The output from the phono amp is very low compared to the line levels
• There are potentiometers by three of the inputs . . . including the phono! (maybe to attenuate the line level stages to match).
• I've never seen a 100K pot across a MM input before - the input impedance isn't close to 47K and the load varies with the setting
• The RECORD outputs are taken from the wrong part of the circuit (should be before tone controls)
• The CUT switch (single pole 70 Hz. rumble filter) is essentially useless.
• The nice Stereo, Reverse, Mono switch is designed to work with the phono pick-up only -- which makes it pretty useless
• There was no phono ground lug.
 

Attachments

  • Leak on bench.jpeg
    Leak on bench.jpeg
    142.5 KB · Views: 87
  • tempImageoKO6hk.png
    tempImageoKO6hk.png
    546.1 KB · Views: 82
  • Leak bottom open.png
    Leak bottom open.png
    730.4 KB · Views: 68
  • Leak top open.png
    Leak top open.png
    663 KB · Views: 77
  • leak_point_one_stereo_sch.pdf_1.jpg
    leak_point_one_stereo_sch.pdf_1.jpg
    288.3 KB · Views: 76
  • Leak Point One Scanned ckt.jpg
    Leak Point One Scanned ckt.jpg
    230.2 KB · Views: 73
  • Replaced parts.jpg
    Replaced parts.jpg
    397.4 KB · Views: 56
  • Leak Ext. PS.png
    Leak Ext. PS.png
    64.6 KB · Views: 53
  • Ext Leak PSjpeg.jpeg
    Ext Leak PSjpeg.jpeg
    204 KB · Views: 83
Last edited:
Register to hide this ad
Initial Listening Tests

After getting everything set up, I connected the pre to my system. There wasn't any hum but the output was way too low to drive my Acro II mono blocks. The phono was weak(er). The tone controls worked. It clearly wouldn't do and more work would be necessary to try to make it acceptable.

Overall, the pre seemed to attenuate the line input signals instead of amplifying them!

Enter (or rather perpetuate) the "sunk cost fallacy." The sunk cost fallacy is a cognitive bias where individuals continue an endeavor—such as projects, investments, or relationships—based on previously invested resources (time, money, or effort) rather than current benefits."

So next I upped the NFB resistor in the line FB circuit from 220K to 440K to give more gain on the line stage, and installed a modified MOSFET buffer on the EF86 outputs.

Now the line outputs were sufficient to drive my amps, but the phono still sucked.

I discovered that the crosstalk was incredible . . . easily the worst I've ever heard. Even a SHORTED line level input bled to all of the other outputs!

So now I've sunk a ton of time and a fair bit of coin with [still] an unacceptable result. So either I give up and strip the unit for parts, or see if I can overcome the problems, perhaps throwing more good time and money after bad.
 

Attachments

  • Leak in rack.jpeg
    Leak in rack.jpeg
    203.1 KB · Views: 37
  • Leak with buffer.png
    Leak with buffer.png
    781.7 KB · Views: 37
Last edited:
Back to the Drawing Board . . .

The crosstalk is related to the high impedance of the circuit overall, making it extra prone to leakage. All of the inputs go to the selector switch, and none of the wires are shielded. Note that they are bundled together . . .

There isn't a practical way to increase the phono output with the existing circuit. Rewiring the selector switch with shielded wires is a PITA and offers no guarantee of a fix. The Stereo / Reverse / etc. switch works only with the pick-up and is an extra place where leakage can occur, so it will be removed from the circuit.

I have small Velleman RIAA circuit that works with 12V. The stock parts are junky and the RIAA is off by 3-4 dB, according to one experimenter who spiced alternative values to give a +/- 0.1 dB result. So I replaced several components with different and better quality parts. I'll go directly from the phono input to the circuit which will give a line level output.

I have a small input relay PCB that handles 4 inputs and runs on 12 VDC. This will obviate the selector switch and keep wire lengths to a minimum.

Wish me luck!
 

Attachments

  • Relay PCB.png
    Relay PCB.png
    1,006.6 KB · Views: 20
  • Velleman RIAA rear.png
    Velleman RIAA rear.png
    982.8 KB · Views: 20
  • Velleman RIA frt.png
    Velleman RIA frt.png
    998.7 KB · Views: 20
This design just seems fundamentally flawed to me if I'm being perfectly honest. Almost seems like it would have negative gain input to output despite having a lot of gain in the tubes. Don't those Leak power amps hit full output at something like 0.2 volts? Tossing out that much signal while also amplifying it just seems like a good way to have poor signal to noise ratio. Leads jammed together like that plus the high impedance / high gain nature is probably why it has the cross talk issues.


it is pretty though. This seems like the kind of unit you want to find that is electronically beyond salvation so you wouldn't have to feel bad about gutting it and rebuilding with something that works better.
 
gadget73 said:
This design just seems fundamentally flawed to me if I'm being perfectly honest. Almost seems like it would have negative gain input to output despite having a lot of gain in the tubes. Don't those Leak power amps hit full output at something like 0.2 volts? Tossing out that much signal while also amplifying it just seems like a good way to have poor signal to noise ratio. Leads jammed together like that plus the high impedance / high gain nature is probably why it has the cross talk issues.


it is pretty though. This seems like the kind of unit you want to find that is electronically beyond salvation so you wouldn't have to feel bad about gutting it and rebuilding with something that works better.
Click to expand...
Yes, I share your concern about the S/N ratio at the end of the experiment. "It would have negative gain input to output despite having a lot of gain in the tubes" is what I experienced in the initial listen.

This is obviously a case where my stubbornness exceeds my intelligence. :banana:
 
Last edited:
Might be blasphemy, but given how nice the cosmetics are, why not drop the guts of a quality preamp (DIY, kit, or otherwise) inside it and just use it for you own enjoyment? I don't believe it's a particularly valuable item to collectors even in great original working condition, so why not?
 
MuseChaser said:
Might be blasphemy, but given how nice the cosmetics are, why not drop the guts of a quality preamp (DIY, kit, or otherwise) inside it and just use it for you own enjoyment? I don't believe it's a particularly valuable item to collectors even in great original working condition, so why not?
Click to expand...
I looked at that too. I have several pre PCBs but fit is a challenge - also have to redesign the ext. PS. Plus there’s the "sunk cost" factors -- the time and materials getting it up and going . . .

I'm hoping I can make a proverbial purse from a sow's ear. If not it's a fail but I've learned a lot.
 
Hard lesson. Reminds me a lot of a project I had with a Heath AA-141 preamp — another miserable design that also ran line level inputs through a portion of the phono preamp section. Luckily, I had no such reservations about quickly yanking the circuit out of it in favor of something much more sensible — and capable. Apparently, Heath was none too attached to the circuit either, and quickly came out with the much revised AA-141A as a result. Would that Leak would have been so proactive!

Dave
 
That's a beauty! A bigger problem than the low sensitivity is that the signal at the output jacks passes through both a balance pot and the volume control. Note that the top of the Balance control is driven through a coupling cap from the plate of an EF86. Even without those controls in the way, this wouldn't drive much of anything. I would build an entirely different circuit into that chassis for compatibility with real-world amplifiers.

Screenshot 2026-05-01 at 8.15.49 AM.png

Jack
 
Are referring to the parallel 100K loads? Yeah thats a bit on the low side for an EF86 to handle. Thats in parallel with 200k worth of resistors that make up the feedback loop for th tone controls.

Also makes for pretty high output impedance since the pots end up adding to the already high output impedance of a pentode. The buffer circuit he added handles that part of it at least.
 
A couple of thoughts . . .

1. I'm going to turn the cut switch into a loudness switch and replace the pot with one with a loudness tap
2. The MOSFET buffer is installed at the output of the pre so that fixes some load issues
3. I plan to replace the 100K balance control with a 250K MN contour blend / balance pot
4. I'm reconfiguring the first tube as a triode stage, which should improve S/N and reduce the need for so much feedback to offset the high gain
 
You must log in or register to reply here.
Back
Top Bottom