This post will cover the design and construction of the phantom power and DC heater filament board for the EMP-351 microphone preamp project I outlined in an earlier post.
This preamp, which is based on a well-known modification from Electric Audio of the repro and record boards in the old Ampex 351 reel-to-reel player will have relatively high gain. In fact, from the strictly hi-fi point of view, it's a bit of a silly design, since it has two full gain stages prior to an unusual phase inverter which also provides significant gain. In the standard mod, this actually necessitates amplification, then attenuation, followed by further amplification, if it is desired to keep the end-to-end gain of the circuit down to the 65-70 dB or typical required for a mic pre. Without the attenuation offered by the calibration potentiometer after the second gain stage, the end-to-end gain of the circuit would theoretically exceed 90 dB -- too much for practical use. Of course, the penalty for breaking the "no attenuation followed by amplification" rule is an increase in noise.
That issue with signal-to-noise ratio is an inevitable compromise that comes along with this kind of circuit. But it is possible to mitigate the problem, at least a bit. For instance, unlike the original unit, the EMP will employ metal film resistors throughout, which are significantly quieter than the carbon composition resistors found in the original. I also made the decision early on that this project should use high quality regulated 12.6VDC for all of the preamp tube filaments, not just the input tubes. In my experience, DC tube heaters are something that should be done right, or not at all. Unfortunately both the original Ampex and the Electric Audio mod rely on a very crude unregulated DC supply consisting of not much more than the rectifier and a big old 4000uF cap. Needless to say, there will be significant residual ripple with even three preamp tubes, and it's unworkable with the six that I need. Nowadays we can do much better.
The six preamp tube that this preamp will have is admittedly a bit of an issue. They'll be drawing a total of 900mA at 12.6VDC. While that might not seem like that much, because of the generally quite poor power factor of low-voltage supplies, it turns out that you actually need quite a bit higher rating (something like 3A, more on this shortly) in order to safely feed those six tubes nice pristine DC.
So things brings us to power transformers. If you think about it, a project like this represents a bit of an unusual situation. Since we have no current thirsty power tubes to feed, the the current requirements for our high-voltage B+ supply are quite modest -- really just a few tens of mA at very most! But the voltages are still in the high range you would normally associate with tubes. On the other hand, we have a disproportionately large thirst for current from the low voltage supply. Unfortunately, most transformers that supply plenty of current for the low voltage secondaries are also designed to provide waaay more than I need for the high voltage. This makes such transformers a bit too large and unwieldy to be practical for a fairly compact rack mountable unit. I was also not too excited about the other options involving two separate transformers. Aside from the space issue, there's the additional unwelcome complication that this unit has to able to work with both 120V/60Hz power and 230V/50Hz.. a consideration that further limited my transformer options.
The logical thing to do, therefore, was to get a transformer wound just for this job, and indeed that's what I did. That said, I also wanted a transformer design that would be flexible enough to use in the future in a variety of small projects, including potentially things involving a low-wattage power tube. In the end, this meant opting for a bit more B+ current capability than I strictly needed for this project. Also, it meant that I wanted the flexibility of being able to power tube heaters from 6.3VAC. So for my low voltage, I actually opted to have two separate 6.3V secondaries with good regulation that I can run in series for 12.6VAC or rectify for powering a 12.6VDC supply (as I will be doing for this project). The trade off here is that I could have made my life a little bit easier with respect to power factor if I had chosen a higher low-voltage secondary (say 15V), but I would have lost the flexibility of using this transformer in other contexts.
I'm having a small batch of the following transformers constructed for me by Heyboer Transformers, located in Grand Haven, MI. It's a pleasure working with the folks at Heyboer, and they come highly recommended:
For one-off or small numbers of boards, I use the Press n' Peel Blue Transfer paper method in conjunction with ferric chloride etching. I wouldn't want to do a dozen boards this way, but it's fine for a couple. Here's a helpful YouTube video made by someone else that demonstrates the method:
This preamp, which is based on a well-known modification from Electric Audio of the repro and record boards in the old Ampex 351 reel-to-reel player will have relatively high gain. In fact, from the strictly hi-fi point of view, it's a bit of a silly design, since it has two full gain stages prior to an unusual phase inverter which also provides significant gain. In the standard mod, this actually necessitates amplification, then attenuation, followed by further amplification, if it is desired to keep the end-to-end gain of the circuit down to the 65-70 dB or typical required for a mic pre. Without the attenuation offered by the calibration potentiometer after the second gain stage, the end-to-end gain of the circuit would theoretically exceed 90 dB -- too much for practical use. Of course, the penalty for breaking the "no attenuation followed by amplification" rule is an increase in noise.
That issue with signal-to-noise ratio is an inevitable compromise that comes along with this kind of circuit. But it is possible to mitigate the problem, at least a bit. For instance, unlike the original unit, the EMP will employ metal film resistors throughout, which are significantly quieter than the carbon composition resistors found in the original. I also made the decision early on that this project should use high quality regulated 12.6VDC for all of the preamp tube filaments, not just the input tubes. In my experience, DC tube heaters are something that should be done right, or not at all. Unfortunately both the original Ampex and the Electric Audio mod rely on a very crude unregulated DC supply consisting of not much more than the rectifier and a big old 4000uF cap. Needless to say, there will be significant residual ripple with even three preamp tubes, and it's unworkable with the six that I need. Nowadays we can do much better.
The six preamp tube that this preamp will have is admittedly a bit of an issue. They'll be drawing a total of 900mA at 12.6VDC. While that might not seem like that much, because of the generally quite poor power factor of low-voltage supplies, it turns out that you actually need quite a bit higher rating (something like 3A, more on this shortly) in order to safely feed those six tubes nice pristine DC.
So things brings us to power transformers. If you think about it, a project like this represents a bit of an unusual situation. Since we have no current thirsty power tubes to feed, the the current requirements for our high-voltage B+ supply are quite modest -- really just a few tens of mA at very most! But the voltages are still in the high range you would normally associate with tubes. On the other hand, we have a disproportionately large thirst for current from the low voltage supply. Unfortunately, most transformers that supply plenty of current for the low voltage secondaries are also designed to provide waaay more than I need for the high voltage. This makes such transformers a bit too large and unwieldy to be practical for a fairly compact rack mountable unit. I was also not too excited about the other options involving two separate transformers. Aside from the space issue, there's the additional unwelcome complication that this unit has to able to work with both 120V/60Hz power and 230V/50Hz.. a consideration that further limited my transformer options.
The logical thing to do, therefore, was to get a transformer wound just for this job, and indeed that's what I did. That said, I also wanted a transformer design that would be flexible enough to use in the future in a variety of small projects, including potentially things involving a low-wattage power tube. In the end, this meant opting for a bit more B+ current capability than I strictly needed for this project. Also, it meant that I wanted the flexibility of being able to power tube heaters from 6.3VAC. So for my low voltage, I actually opted to have two separate 6.3V secondaries with good regulation that I can run in series for 12.6VAC or rectify for powering a 12.6VDC supply (as I will be doing for this project). The trade off here is that I could have made my life a little bit easier with respect to power factor if I had chosen a higher low-voltage secondary (say 15V), but I would have lost the flexibility of using this transformer in other contexts.
I'm having a small batch of the following transformers constructed for me by Heyboer Transformers, located in Grand Haven, MI. It's a pleasure working with the folks at Heyboer, and they come highly recommended:
Primary 115/230V
Secondary # 1 240 Volts
@ 115 mAmps
Secondary # 2 6.3 Volts
@ 3.0 Amps
Secondary # 3 6.3 Volts
@ 3.0 Amps
Drop me a line if you are interested in one of these transformers, I'll have at least a couple left from this order that I don't have immediate plans for.
Before we go any further, here's a link to the (PDF format) schematic for the filament and phantom board. As you can see from the schematic, this was actually drawn in LTSpice, my favourite (and free!) circuit simulator software. I have, in fact, confirmed the proper operation of this circuit with the simulator. NOTE: a fuse is not shown on this, but is obligatory for any real-world implementation.
A couple of points.. the phantom power is derived from a fairly standard voltage quadrupler circuit. In principle, I suppose I could have used a voltage tripler, and referenced it to the output of the main rectifier. But the inherently half-wave operation of a tripler wasn't appealing, both in terms of the unbalanced current established in the transformer, and in terms of the limited available current. The choice of the TL783C adjustable high voltage regulator, was influenced by Douglas Self's recommendation of this part in his excellent book "Small Signal Audio Design". Every serious audio electronics hobbyist or audio professional should own this book.
As to the heater filaments portion of the design, it's fairly standard, and follows the treatment given by Merlin Blencowe in his excellent (but sadly now out-of-print) book on tube amp power supply design. This will be the first time I've used the Micrel MIC29300-12WT regulator, which is a promising looking high current, low dropout part that is newly available from Mouser. Unlike the TL783, the MIC29400-12WT will be dissipating 3W or so of power, so will be heat-sinked to the aluminum chassis.
Now, of course, a schematic only gets you so far. Long time readers of this blog know that I'm a big fan of the freeware program DIYLC (do-it-yourself layout creator). I do most of my layouts, both for hand-etched PCBs and for entire large amps, using DIYLC. Because of the regulator chips and the small adjustment pot, it made a lot more sense to design this one as a PCB rather than as a turret board. So here's what the PCB layout looked like once it was implemented in DIYLC:
Here's what the mask looks like with no components. Resize this image to 4.75" x 3.75" if you wish to etch your own board based on this. Incidentally, for anyone who wants to build this, I strongly recommend that you chose 105degC rated low-impedance ("low-Z") capacitors for the large electrolytic (1000uF and 4700uF) shown on the schematic.
For one-off or small numbers of boards, I use the Press n' Peel Blue Transfer paper method in conjunction with ferric chloride etching. I wouldn't want to do a dozen boards this way, but it's fine for a couple. Here's a helpful YouTube video made by someone else that demonstrates the method:
Here's what my board looked like at the various stages of this method. First is with the Press-'n-Peel ironed on:
Next, we dump it in the ferric chloride solution and swirl it around for forty minutes or so until all of the exposed copper is etched off:
After it comes out, it'll look this initially. Note that the Press-'n-Peel is still stuck to the copper areas:
This needs to be gently scrubbed off to reveal the copper traces:
You can stop here, but I find that soldering is made much easier if the copper is first coated with a layer of tin. I do this by an application of a product called "Liquid Tin". It's easy to use and works great:
At this point, you are ready to drill the board and mount the components! Once again, here's what the finished board looks like!
Everything here is attached except for the 12V regulator, which as I mentioned will be heat sinked to the aluminum chassis when the board is installed. Needless to say, it's extremely satisfying to start with just a concept, and end up with its physical realization in your hand.
Stay tuned, next time we'll begin looking at the main circuit, and follow along the construction of one of the main turret boards!
No comments:
Post a Comment