|
|
| Mark's Project Pages/Audio Projects/Hi-Fi Preamp/Initial Plans/Log | |
|
Hi-Fi Preamp - Log: Introduction: Wow! Looking back at this stuff, I clearly had far too much spare time! Here's a html'd version of a Word document I maintained while planning version 1 of the pre-amp. How's the insomnia?!
February 1, 1998 Initial Specification: User Features:
Technical Details:
Construction Details:
Notes: The physical size is currently arbitrary, based loosely on a half-suitable box I already have. It is hoped that the power-amp will be able to use complimentary dimensions (it'll probably have to be taller than the pre-amp, of course). I've selected some switches to use - the MEC UNIMEC range of DPDT push-button switches. I'm recycling the switches from old panels, but I've already bought some shiny new key-caps, ready for completion. The switches work with a metallic click, and I'll need to watch that - they can sound a little 'cheap' on an undamped thin PCB. The mounting pitch is ½", defining the size of the unit, as described below: Front Panel Layout: This is an initial guess at the front panel control layout:
The volume control would appear at the right-hand side, the switches underneath. Clearly, this is far from scale! The Standby switch is on its own, the Dubbing and Record Out switches are paired together, and the 7 sources switches are in a group. The spacing between groups is that of a switch, making a total width of 6". Allowing 12.5mm either end of the switches makes for a window of 178 mm. Adding 30mm for the volume control immediately puts the width up from the 200mm suggested previously. At this stage I suggest 230mm. "User Interface": Simple - press switch for desired source. Can't be easier. The LED's above the (green) switch legends will indicate the selected source (in yellow), and the top row of LED's will indicate any record selections (in red). The Standby function kills all those LED's (lighting a red one above the switch) and mutes the output. To suit purists, I'm probably going to leave the analogue stages powered up during standby. I have a choice about the record-select relays - should I leave them engaged during standby, thus preventing recordings from being interrupted? To select a source to record, press Record-Out. Its LED will light, also the legend back-lighting will change to reflect the change of function of the source switches: Tape 1 and Tape 2 will go out, and the remaining 5 will change from green to red. To cancel this mode, Record-Out can be pressed again, or it will 'time-out' after a few seconds. If a source button other than tape 1 or 2 is pressed, the record-out relays route the correct source to the tape-out relays, and the top row of LED's echo this selection. If one of the tape inputs is pressed, no source is routed out - this can be useful to prevent any source-loading effects… To dub from one tape-loop to another, the Dubbing button cycles through Dub Tape 1 to Tape 2, Dub Tape 2 to Tape 1 and no dubbing. When dubbing is enabled, the LED above the dubbing switch lights - also a legend lights in the top row of LED's (above Tape 1 and Tape 2). Any previously selected Record-Out source still remains valid for the appropriate tape loop - i.e. if CD was selected as record-out, and dubbing from Tape 1 to Tape 2 is selected, then Tape 1 still receives CD as it's input. The remote-control should have a pair of buttons for volume, a standby button, and 7 source switches. I don't expect to control the record-out/dubbing functions from the remote (?). Also, a mute function would be easy to include, but another button on the front panel would add an inch to the width (unless it was pair with the Standby switch). This could use the existing output-mute relay, or more-usefully it could be a -20dB function - bearing in mind how pots are often bad at near-minimum settings - also a motorised pot might be tricky to control at low settings…
March 23, 1998 After some initial research and attempts at PCB layout, some changes to the specification have come about… I spent some time trying to design a suitable motherboard, that could support all the sub-boards while offering an 'electrically good' layout, and came to the conclusion that it was easier to incorporate the relays directly on the main board. Also, another factor influenced this decision: research suggests that poor results would be obtained using CMOS technology, hence the ability to experiment with CMOS was possibly not all that useful after all:
Rather than having a motherboard with a large 'bus' carrying the signals from the rear of the PCB underneath the two relay boards, I have decided to place the relays (for listen and record) 'facing' each other, so that the signal from the phono sockets can connect to each relay via less than ½" of PCB track, Molex connectors, and a short length of twisted-pair cable. Another thought that has come out of the experiments - I was planning to incorporate the relay-drivers and decoding logic on the relay sub-panel (because if I had used CMOS switches - i.e. 4051, they would require the 3-bit binary code directly). However, I have decided to move the drivers to the control panel - mainly for ease of layout. This has a few advantages - for example I could produce a non-remote version of the product, which uses simple rotary switches for selection… (or indeed a test-panel for development purposes). I feel that it is best to keep the analogue motherboard as simple as possible, so that the layout can be optimised for signal performance. In the space that I have allowed, I'm having difficulty fitting the phono pre-amp on the motherboard. (I'm aiming for a width of 230mm, the toroidal and power supply will live on one side of a partition, placed 80mm from the left of the case, leaving 150mm for the motherboard). I may end up mounting it vertically against the side of the case, which limits it's width to 50mm, although it will have quite a lot of length… My current motherboard design currently has a Molex connector for the phono source, like all the others. Perhaps the phono pre-amp could be an optional extra, as fitting would involve unplugging the Molex, attaching it to the phono pre-amp board, and then plugging the phono pre-amp to the motherboard… Line-Level Amp: I'm considering using a discrete design, based around a symmetrical topology, as used by the A1 and other well-known designs. I think I could use 2N2905's and 2N2219's as output devices, and bias them into class A. I've found some large TO5 heatsinks that would 'look the part'… I reckon that it would be useful to have a 50ohm output impedance, to negate the effects of cable capacitance. I think that a good, low-noise, low-distortion, wide-bandwidth design could be arrived at using minimal active devices, and what with lots of local negative feedback, I think it could appeal to the 'hi-fi' community! Phono Pre-amp: I have found a datasheet for the NE5532 which gives a very high specification RIAA pre-amp. As I'm not really in a position to test for the ultimate in sonic design with my record-playing kit, I suspect I will use a 'close relation' of the given circuit. It has to be better than what I'm currently using… I think that although the design uses 2 op-amps per channel (that's for MM only), I should be able to fit it in the available space (remember - only 50mm of width - an inch per channel)
March 31, 1998
Given some more thought, it might be better mount the mains TX on the right of the case, enabling the control and PSU PCB's to be closer together, as the ribbon cable will otherwise have to be routed above the TX, causing noise (?) problems and untidiness… The problem with that is the volume control shaft (not shown in the diagram, but the pot is mounted on the right hand side of the partition - as the RIAA PCB is drawn, it interferes with it). However, it's worth some thought, as that might produce a more logical layout - taken to it's logical conclusion would result in the RIAA pre-amp being mounted on the left, with the mains input on the right. Also, the ribbon-cable to the front panel appears 30mm from the left, so that would fit better… I'm currently not too worried about space for the power supply or control PCB's as these can have a tight layout - I want to give as much priority to the signal stages at this point. I suspect I might move the mains TX to the right, and live with a longer run of mains cable from back to front. There's always the out-board PSU option, of course!
April 19, 1998 Another thought - I'm considering separating the relay panel from the line-amp, or in other words abandoning the idea of having the line-amp as a daughter-board on the relay panel. The only drawback with this is the increase in internal wiring, but I think there are lots of advantages, most obviously a cost saving - a large area of empty PCB seems rather wasteful. A way to help keep the wiring tidy would be to maintain the main ribbon-cable 'bus' to the relay panel, and then power-supply connections can link from this board. Another way would be to split the ribbon-cable into several feeds for each board - this might appear to have better 'signal integrity' but might make servicing more difficult (currently, in the event of a fault, (almost) the entire cable harness can be checked by substitution using a standard IDE cable). However, splitting ribbon-cables is a known technique for making them more manageable - for example a 40 way ribbon-cable is roughly 60mm wide - meaning that with our maximum height of 50mm, we cannot otherwise route the cable along a vertical path. I'm fairly relaxed about the signal connections - I think that using twisted pairs from the relay panel to the line-amp (via the volume control) will not compromise performance greatly, and will make construction somewhat easier by relieving the difficulty of achieving the accurate alignment required for a PCB-mounted control.
June 19, 1998 After a long break (due to work), I've finally made some progress. After spending lots of time worrying about space, and trying to decide on a size that wouldn't cause difficulty, I've taken the plunge and fixed a case size. The problem with trying to plan to the nth degree is that nothing gets done! After clearing out the attic at work, I've acquired lots of aluminium extrusions and panels, and also several top/bottom panels from cart machines. These are U-shaped panels of 1mm aluminium that measure 285x218 (depth by width) and there are 2 different heights: 44 and 47mm. So, using these as a top panel wrap gave me a starting point. I then selected some ideal extrusion that formed the internal side panels (although these would look fine if they had to be on display - for example I could choose to use a tinted glass or exotic hardwood top panel instead of the wrap). Finally, I remembered that I had 4 large pieces of 2.5mm aluminium sheet that were discarded back panels from a kit that TRL used. After lots of drilling and filling, I now have the basic case, which is uncannily close to the original specification:
I've made the front panel somewhat wider than the case to give it the mini-rack case 'look', although I could decide to trim the ends down later, perhaps… The only downside is that the display window that I had designed earlier is a bit too wide, which means that I'm having to rethink things slightly - perhaps there will be an interface re-design…
July 6, 1998 I've done some PCB design, and thought some more about the interface. I'm going to drop the dubbing button (which only saves ½" but I think that's enough). Now, when one presses the RECORD/DUBBING switch, you can press TAPE1 to dub from tape 1 to tape 2 (i.e. tape 1 is the source). Pressing TAPE2 lets you do the opposite. Previously, pressing TAPE1 or TAPE2 muted the record-output, so now you can do that by holding the RECORD/DUBBING button for a 2 second period. I think that this shouldn't cancel the dubbing selection - I guess the best way to do that would be to press RECORD/DUBBING, followed by the tape source that is already selected (i.e. toggle it off) I've designed a RIAA PCB (based on the NE5532 application sheet) that measures only 38 by 120mm. I've also managed to fit the PSU onto a 35 by 115mm board. The latter was especially hard work because as well as 6 separate regulated outputs, I've included a power-good monitoring circuit, which triggers a time-delay start-up on the relay supply. The power-good signal is made available on the power ribbon-cable (there are 3 separate ribbons in this design - the power cable visits all boards except the front panel, the relay cable travels between the input relay board and control board, and finally there is a wider cable that links the front panel to the control board. Obviously, much consideration has to be made to ensure neatness and serviceability) The power-good signal is bi-directional in that the control circuit could take it low during Standby (which will kill the supply to the relays) and then make use of the supply delay during start-up. Of course, that would kill any record-selections made, which may not be desirable. That is an important consideration to remember when experimenting with component values - if the power-good detector is too sensitive, playback and recording would be interrupted needlessly - I guess the actual amplifying circuits would hold-up for some seconds… perhaps I should monitor the 5V rail rather than the AC input - after all the point of the power-good is more to provide a clean reset for the control circuit… Physically, the PSU uses the inner partition for heatsinking and support. To make servicing (and PCB layout) easier, I decided to space the regulators 3mm in from the edge of the PCB, and use a length of 3mm aluminium as a spacer between the devices and the partition. Therefore, the 6 devices can be bolted to the spacer, and the spacer can attach to the partition with 2 or 3 screws… I need to take a few millimetres off the width of the relay input selection panel, but once that is done, it will fit as the plan shows. I've obtained a motorised potentiometer, which is much bigger than I expected, so this means having to intrude in the space allocated to the control panel. I have been gently investigating the use of a PIC to perform the bulk of the logic and decode the remote control for me. This, in theory at least, should reduce the component count and complexity, albeit at the expense of development time… I could make use of that remote control that didn't go back to CPC (the Philips satellite receiver handset). I've seen a data sheet for a Philips IC that would decode the RC5 for me, but that seems somewhat hard to find… I can see the other potential problem with using a micro-controller - I would be tempted to incorporate features such as an LCD display and menu operation… all a bit OTT really. That said, I don't think a PIC would make that easy for me - having seen how difficult they are to program!
July 28, 1998 Again, not much progress, due to work. In addition to researching PIC programming, I've obtained a Motorola 68HC11 reference handbook. This device seems much easier to program, as it shares many instructions with the 6502 that I know and love. This device is programmable via a serial port, which makes development and firmware upgrades much easier. I guess I could decode RC5 in software rather than using the SAA3049 which, as mentioned above, is very hard to obtain. I've polished the PCB layouts, by redesigning my component libraries, and making sure things will fit. This was important for the PSU, as things are very tight. At the moment, I'm having to modify some of the values used in the phono pre-amp, as some of the capacitors are awkward values. Until I've made some decisions about the control panel, front panel and line amp, I'm stuck. I guess that I could initially omit the line amp, but that might mean that the thing will go into service and never get finished…
Things stopped here due to the small matter of a house purchase, amonst other things...
|
©2004 Mark Hennessy
Contact Me | Site Map
| Disclaimer
After
some more thought, I've had an idea that may solve the space problem: mount
the toroidal transformer at the front of the case. The problem was discussed
above - 80mm is needed by the transformer, 140mm for the main board, not leaving
much from 230mm for the phono pre-amp. If I put the transformer at the front
of the case, the power supply can live in that side-gap, leaving more space
available… (see diagram)