New 990 Discrete Opamp - 100% SMD version

elskardio · Mar 10, 2020

Hi Guys,

I wanted to share my latest version of the 990 discrete opamp.

It's 100% SMD... but why? ???
I've been building a lot of 990s lately for my Twin Servo preamp and it's pretty time-consuming. So how accelerate the assembling process without increasing the cost? SMD to the rescue! I can simultaneously build a panel of multiple opamps and a lot faster than my through hole design. No need to bend component legs or flip the pcb to solder.

All resistors are MELF. Capacitors are COG with the exact values from the original JE-990 schematic. My main concern was the heat dissipation of the output transistors. For most application, I need the 990 to be stable at +/- 16V driving a load of 600 ohm. I just finished stress testing a pair running at +/- 18V driving 150 ohm for 48 hours... I can keep my finger on the transistors

Now the big question... how does it sound? I only did a quick test between the through hole version and the SMD and I couldn't tell them apart. I sent a pair of Twin Servo to a local studio yesterday. One with a pair of the 990 SMD version and the other with the through hole version. I'll have the audio files next week to compare. If you guys are interested, I could post the files when they're available.

Cheers
Gabriel

Here's a picture:

The output transistors are mounted on the bottom.

I love how low-profile the SMD version is...

fazer · Mar 10, 2020

Lean design. Nice. Do you sell these or are they just for your use?

JohnRoberts · Mar 10, 2020

I seem to recall Paul Wolff converting his API DOAs over to SMD last century...

JR

fazer · Mar 10, 2020

Those api amps do have a hi failure rate.

john12ax7 · Mar 10, 2020

How many MELFS rolled away in the process of assembly?

Good idea with SMT, I've heard multiple people say its faster to hand solder than thru hole.

Fuzz Face · Mar 10, 2020

Really nice looking work as usual. I like how you thermally coupled the diodes to the output transistors, very clever.

Did you do all soldering yourself or are you able to order them with some parts already soldered on?

I want some, if only just for fun

pucho812 · Mar 10, 2020

JohnRoberts said:
I seem to recall Paul Wolff converting his API DOAs over to SMD last century...

JR

yes. and there are some people from last century who claim to hear a sonic difference between the smd vs thru-hole. I don't by it but they claim it.

elskardio · Mar 10, 2020

john12ax7 said:
How many MELFS rolled away in the process of assembly?

1 or 2 haha!

I pick them from the tray with a pair of tweezers and tack directly one side to the PCB.

fazer said:
Lean design. Nice. Do you sell these or are they just for your use?

I'll have the first batch of PCBs in 2-3 weeks. If some members are interested I guess I could sell a few. All the parts are available from Mouser or Digikey.

Fuzz Face said:
Did you do all soldering yourself or are you able to order them with some parts already soldered on?

All parts are soldered by hand.
Eventually I could have theses fully assembled by the PCB Fab house in China... but for a small run, it's overkill and out of budget.

JohnRoberts · Mar 10, 2020

pucho812 said:
yes. and there are some people from last century who claim to hear a sonic difference between the smd vs thru-hole. I don't by it but they claim it.

:

I generally found that SMD did better than similar TH designs... same exact parts in smaller packages, work pretty much the same (or better). :

I found benefits from tighter component layouts with less loop area for magnetic induced noise pick up. Not sure how much of a factor that is for DOAs...(it mattered in mixer mic preamps done by my engineers, last century. )

In my judgement Paul was a great marketer and serviceable design engineer. I could never schmooze customers as well as he did.

JR

Winston OBoogie · Mar 11, 2020

Fuzz Face said:
... I like how you thermally coupled the diodes to the output transistors...

+1. Nice layout and soldering dude

evil grill · Mar 11, 2020

JohnRoberts said:
:

I generally found that SMD did better than similar TH designs... same exact parts in smaller packages, work pretty much the same (or better). : I found benefits from tighter component layouts with less loop area for magnetic induced noise pick up. Not sure how much of a factor that is for DOAs...(it mattered in mixer mic preamps done by my engineers, last century. )

In my judgement Paul was a great marketer and serviceable design engineer. I could never schmooze customers as well as he did.

JR

Interesting. I immediately came to think of this: www.davehilldesigns.com/smt_resistror_distortion_rev1.pdf

Tekay · Mar 11, 2020

What did you use instead of the LM394 on the input?

gyraf · Mar 11, 2020

..beautiful work, that..

living sounds · Mar 11, 2020

Looks great! How about a DIP8 version? 8)

Dan Kennedy · Mar 11, 2020

Very nice work. I'm guessing SSM2212 input pair?

JohnRoberts · Mar 11, 2020

evil grill said:
Interesting. I immediately came to think of this: www.davehilldesigns.com/smt_resistror_distortion_rev1.pdf

The semiconductors are generally the exact same parts just in different packages.

"Small" resistors can be problematic even in through hole packages. The problem with small resistors expresses with higher voltages. I've seen compromised linearity (measurable distortion) in audio power amplifier feedback networks that can have tens of AC volts across them.

More recently when I was designing my outlet tester design, that used smd parts, I found that to realize high resistances at high voltage (tested at 500V) I had to avoid using too small SMD resistor packages. [edit- I used a 1206 for the 100M resistor /edit[

In my judgement it is not a SMD issue but a size issue, while I doubt you will ever find a leaded through hole resistor as tiny as an 0204 smd.

Considering all these sundry factors is what design engineers do...

JR

elskardio · Mar 11, 2020

Winston O'Boogie said:
+1. Nice layout and soldering dude

gyraf said:
..beautiful work, that..

Thank you guys

------------------------------------------------------

Tekay said:
What did you use instead of the LM394 on the input?

Dan Kennedy said:
Very nice work. I'm guessing SSM2212 input pair?

Yes SSM2212 on the input. I guess it's the official replacement now... John Hardy is using the same.

------------------------------------------------------

living sounds said:
Looks great! How about a DIP8 version? 8)

hummmm... Why not

------------------------------------------------------

evil grill said:
Interesting. I immediately came to think of this: www.davehilldesigns.com/smt_resistror_distortion_rev1.pdf

Thanks for sharing that. I remember reading a comment from a wise guru somewhere (can't remember who) about choosing SMD resistors for audio. His conclusion was, choosing thin films with the lowest temperature coefficient in size 0805 or bigger yielded to lower distortion. If you need high power rating, you parallel multiple resistors.

moamps · Mar 11, 2020

Nice design, but I will suggest for next version moving output transistors on top and soldering small heatsinks around.

elskardio · Mar 21, 2020

Hi Guys,

next week I'm assembling a full panel of the latest revision of my SMD 990.

If you want one, PM me. $30 USD each + shipping.

Cheers

elskardio · Mar 23, 2020

80hinhiding said:
Can your variation on the 990 drive 75ohm at +/- 24VDC?

No SMD transistor will tolerate that amount of heat!
As mentioned in my first post, my SMD version will happily drive 150ohm at +/- 18VDC.

New 990 Discrete Opamp - 100% SMD version

Help Support GroupDIY Audio Forum:

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Similar threads