New "Kit" on the Block - BrianGT /LM3886 Upgrade
diyldr – AKA Uriah Dailey – has been teasing me for months with information on a newly developed LDR based mod for the well-known and highly regarded BrianGt amp from Chipamp.com. It simply replaces the input resistor and is quite easy to build and install.
Attachment 268841 Attachment 268842
The PCB measures just 23 x 8 mm and uses 7 components to create a tiny power supply on a tiny board with pads to attach to the LED legs of the LDR. The size of the unit allows the builder to install it in a variety of positions related to how the BGT amp is mounted. In these pictures short leads was the goal.
Attachment 268843 Attachment 268844
Here’s a link to an information page - https://picasaweb.google.com/udailey/ResistorReplacer. There are some more pictures of the modules I built as well as a hook-up template.
Maybe the best introduction is just to relate what I did Saturdau morning. I finally got to a place where I could do some quick exchanges in an effort to compare several projects.
Here is my platform.
Source: Dogbreath modded PS-1
ICs: Boutique 1 foot shielded w/gold RCAs
Pre: JC-2 kit from Jim’s Audio – eBay.
ICs: DIY single strand solid silver (30 ga.) unshielded w/gold RCAs.
1. Hafler DH-120 – stock
2. Hafler DH-120- w/ Power Cap and board upgrades.
a. Hafler DH-120 Power Amplifier Upgrade Kit | eBay
b. Hafler DH-120 Power Supply Capacitors | eBay
3. BrianGT Dual Mono kit w/U. Dailey LDR Resistor Replacers
4. MyRef-C w/ LF01, Mundorf Power caps, Sonicap and Blackgate mods.
I have two others – Carver m1.0t and BPA-150s from Jim’s Audio that weren’t included this time.
I used two tracks from a single CD - Philips Digital “Best of Saint-Saens”. Track (1) has a lot of solo string segments mixed with pizzicato, woodwind, brass and percussion embellishments. Track (6) is a soaring orchestral finale complete with a full pipe organ and a bass drum that sounds like it has about a 12 foot diameter. Between the two tracks one can hear almost everything a system could be asked to produce. I left the pot on the JC-2 in the same spot for all tests.
The stock Hafler produced the largest bottom end of all the amps but did so with a slight lack of control. Some distortion/garbling in fast moving bass and low brass passages was evident and the dynamic mid and highs had a shade of glare in the loud parts. Though enjoyable to listen to, there was very little depth in the image and some noticeable horizontal frequency dependent shift between speakers.
The modified Hafler was cleaner/smother and had more control over the low end. Instrument definition and separation was better, but still little depth. The stage was a little wider and the glare was tamed.
The BrianGTs sound like a completely different amp with Uriah’s new module. They have long been my second favorite, and if I had never built anything else I would have been quite satisfied with them as my system core. Only after hearing the MyRefs did I notice they were a little forward/pushy. On some jazz and pop music, that characteristic can be very enjoyable to my ears. IMHO, I hear the stock version of this amp to be in the same family as BP-150s as great devices for simple builds and high quality listening.
Initially I installed the RR in only one amp for A/B comparisons. The clarity and detail improvements were outstanding. As with the two Haflers, the modded BrianGT exhibited what initially sounded like a bit less low end, but with more time I understood it was actually better balanced and controlled. To investigate any SPL differences I used my Radio Shack level meter with the second track of the test CD. Both peaked at 84.
After installing the RR in the second BGT, all the promise of a new and improved presentation was realized. Just about all the elements – clarity, definition, depth, width and balance could be heard. One would not expect such a dramatic improvement (to my ears) from such a simple, elegant and easy to build modification. I still want to push the new setup with the LighterNote, but I’m currently in the process of re-calibrating it.
To clear up any mystery the MyRefs came out on top, as usual. (Sorry diyldr :cuss: :D). The word “Reference” IMHO, truly describes what the MyRef-C and its derivatives are capable of. As I have noted often, what I hear when using that design can most accurately be called “letting the music come through unobstructed”. Every mod I have added seems to simply remove some element that clouds and/or clutters the sound. There are several builders who are deeply involved in tweaking the MyRef to squeeze out everything possible. Dario Inserra has also developed a major rework that promises even more. http://www.diyaudio.com/forums/chip-...ne-tuning.html
I won’t wax glorious about my go to amp cause that’s not the focus of this post/thread. Much has been written and I’m going to side-step by just saying – I encourage folks to build one or find an owner near you to audition the sound produced.
I would suggest Uriah’s LDR Resistor Replacer does for the BrianGt what Siva’s LF01 does for the MyRef. Uriah is still developing these modules and I suggested the only thing I would want is a touch more mid frequency energy. There is no doubt the concept Uriah uses here can and should be explored for use with a variety of power amps – maybe even the MyRefs. The RR allows the BGTs to produce a lot of the refined character of the MyRefs, but to my ears is a somewhat dryer sound. That’s not a criticism – just a personal preference. The depth and width of the soundstage beats the Haflers hands down, but is just short of what the MyRefs can do.
The module is available in both kit and completed forms on Uriah’s web site. Build An Amp
Kudos to Uriah for developing a super economical yet dramatic design improvement/addition that should be of high interest to all BrianGT owners.
Hopefully, this thread will attract lots of BGT veterans. I’m sure Uriah will welcome and encourage feedback from users to aid in this development process.
Forgive me for having a headache....but, what exactly does this mod do, in terms of the topology and schematic for the amp? I understand the mod permits the user to vary the value of the input resistor....but why/how does this improve the amp from a sonic perspective? I'm lost. And my head hurts......(!)
Bob, thanks for linking me to the thread and thanks for putting it up.
CanAm Man, LDRs just sound better than other resistors. Its not about varying the resistance its about replacing the resistor with an LDR. If you read the Lightspeed or Warpspeed or DCB1 or ZenMod's LDR B1 threads you will find a lot of folks who hear a sonic improvement with LDRs replacing potentiometers. I figured thats just a variable resistor so lets try replacing regular resistors with them. In my tests they sounded markedly better than any of the high end resistors I bought to compare them with. I put them in my BrianGT chipamp and the improvement was significant. I sent them to Bob to see if he heard the same thing and this thread is a summation of what he heard.
The schematic of the amp doesnt change at all. We still simply have a 1k resistance on the input. This time instead of a Dale or whatever, its an LDR. Silonex NSL32SR2.
I posted the schematic in my Picasa album that Bob linked to. Feel free to build one up and try it. If you decide to do that I will help you with any questions. Would love to hear your impressions.
CanAm Man, So sorry to hear about your head :spin:
The post may well be a case of TMI. I took that path just because that morning session was simply the first opportunity I had to do a bunch of comparisons on several projects I've been working on. I presumed potential users would separate out (as you did) the specific LDR RR chunks that could lead to an active and informative discussion.
As is always the case - hearing a modification first hand in one's own system is the best approach. The cost and simplicity of this one hopefully will allow some fun and productive results for many.
Thanks, guys. I had heard that about LDRs--having better sonic qualities.
I've yet to figure out why.....but I guess I'll keep digging into the research and theories.
This is an interesting use of LDRs, thanks for posting it. In passive attenuators such as the Lighspeed the LDRs are used to replace a potentiometer for the following reasons:
a. Potentiometers generally have poor tolerances.
b. High quality potentiometers cost a good amount of money.
c. Potentiometers include moving parts that can oxidize over time causing connection issues, some people theorize this leads to a rectification effect.
In your use you are replacing a precision resistor that has none of the above issues, it's very interesting that you report an improvement in sound quality. Any thoughts on what characteristics of the LDRs make them sound better?
I hadn't run across the LM334 before, might have to buy a couple of those to play with some LDRs I have sitting around.
An alternative usage for the LDR could be the convenience of variable input impedance, since not all sources have the same requirements.
It is also possible to drive an LDR instead of the LED in a clipping detector or distortion detector circuit. This would be extremely useful considering the counterproductive clipping performance of spike system chips. Instead of allowing the chip onboard limiter's terrible noise, the LDR would have turned down the volume just in time to avoid the spike system's typical screech. Although a light sensitive transistor based optocoupler is technically better for that because it reacts instantly, the slower LDR might be more dynamic for having allowed a tiny timeframe for full blast prior to cutting the noise. Both are popular in prosound equipment and is actually a form of compression, but that sort of circuit does compression instead of clipping, which is a desirable trade off.
Most chipamps seem to use 680R in this position. I take it there's no problem adjusting this to give a higher resistance to drop the gain a bit, say 1.2k or 1.6k?
You could check DC offset before and after. You may like adjusting the feedback shunt resistor (680R) to some other figure -and/or- you may like adjusting the feedback resistor (22k) to some other figure. Either and/or both can adjust gain, can make different dc offset results (or different dynamics results) and can make different sonic signature.
Before you start, most sources and your amplifier will appreciate it if you change the 22k input load to 10k or 11k because higher current there (in+ to 0v) will make available many more gain divider (feedback and feedback shunt) resistor combinations result in zero offset.
Before you start, you probably want to install 220pF from IN+ to IN- directly at those pins of the chip to assist stability and block RF gain.
Disconnect the speaker (or temporarily use a large series cap to block DC)
Short the NFB cap temporarily (for test purposes)
Try several gain divider resistor settings
Use the lowest offset (measured at speaker jack)
The amplifier has now told you which settings it likes (lowest offset)
Remove the short from the NFB cap (restore normal function)
Reconnect the speaker
If there are several possible gain settings that also result in zero offset, you could decide practically, by listening if you like. There are several ways to check stability and it is always good to check that. If you lack a laboratory, you can choose options that result in simultaneously best sounding with coolest running as that combination is a good (but not certain) indicator of either stability or decreased workload.
NFB Cap size. The chipamp.com manual has the wrong size cap and the chip datasheet is not better as neither is big enough to assure a level response with electrolytic capacitors. The point is big enough and also clear, thus you can use the right size and/or any larger size.
After you're done setting the gain, review the NFB cap + feedback shunt resistor RC size to insure that this high pass filter allows all of the audio band (esp bass) to pass and not accidentally a bass blocker.
If you favor an easy practical test, you can put an additional much larger cap parallel with the kit's cap and simply listen. There is much variety possible, much of which is favorable. A pretty selection is better and safer than omission, and the amplifier (with the NFB cap) usually lasts longer.
Gain? If you already have a preamp that you really like (or a super-loud source), then decreasing power amp gain is probably beneficial since that will also decrease the influence of the power amp and thus lower gain seems a good consideration for Spike system chips like the LM3886.
Output RC (zobel). The chipamp.com kit has used a more decorative and mistakenly more efficient cap for speaker zobel. The chip datasheet by listing 2.7R assumes a lossy cheap polyester dip cap, which is the normal part for production amplifiers. Those low price dip caps have about an additional 4 ohms worth of loss inside. The actual resistance value should be 6.7R, same as the DCR of an 8 ohm speaker. You may want to change either the resistor or the cap so as to upgrade to normal component values. However, even smaller value caps (such as ~22nF dip cap) still work much better than omission.
There's really a lot of little errors to fix on the chipamp.com amp, and it would be useful to do so before optimizing gain settings. One possible advantage of the aftermarket upgrade kit (on THIS thread--see post#1) is in lifting some of the small signal components up off of the noisy little board. Certainly an additional upgrade to Signal Star ground could be a worthwhile consideration.
Daniel could you post the link you mentioned in your last post? It didnt stick on your last one.
Good idea to use as a variable input impedance. I like that. At this low input impedance every few hundred ohms is at least 1% change. A total 22k impedance could change to 32k and thats substantial. The LDRs can go to 5k-10k without moving around like crazy but after that I dont trust them. You can get some that will even be kind of stable at 15k. I never have had any that behaved after 22k so I use them in positions of a few k or a few hundred ohms and they give me a great boost in sonics and I can count on them to behave as they should.
|All times are GMT. The time now is 08:04 PM.|
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio