还是六根不净啊，没能抵挡住诱惑哦．．．

longshort

这一向一直在找一种价格和性能都能够承受的高精度运放，准备用于未来的电压参考源中。其实自己手中本来有几个国产的低漂移运放FC74，查参数也不错，失调温漂在0.5uV/摄氏度，但一则是体积稍大，同时使用几个的话均温不易，而且这几块电路都是当年直接从生产线上撤下，连标签都没打印，因此是否老化、参数是否超标都不知道，还是不太敢用。

于是又上网找合适的，但是价格都太贵了。有些老型号罐形封装的竟然要两百多，实在太吓人了。而一般的高精度、低漂移运放大多在二十多到五十多之间，价格太低的实在不敢相信，心里想想要实在不行，以后再说吧。

这当儿又在自己常买东西的一家淘宝店里看到几种运放，带前缀“OPA”的有好几种，其中OPA227是高精度运放，但其他几种没有资料不知道，于是再展开一场搜索资料的行动，终于发现OPA735这一型号好生了得！查其datasheet，失调电压最大为5uV，这与其他高精度运放相当，但最大失调电压温漂竟达0.05uV/摄氏度，而长期温度稳定性数据，在300小时150摄氏度的寿命测试中只有1uV的散布！这实在是太优秀了！

踌躇了好些日子，翻来覆去地将其与OPA2227比较，除了2227价格为12元之外，2227的封装形式也对精度有影响，而2735无论是单还是双，精度都保持一致。但是2735要卖18大元哪，心疼啊！最后没辙了，谁让俺想她来着？掏兜儿吧．．．

不过想来想去，总觉得这玩艺儿要达到这么高的温漂指标，不用点儿特殊技术是很难办到滴。也许这就是斩波稳零运放？datasheet里头没提供等效电路图，这又是种低功耗器件，水实在深呀！！！

刚才那家网店的老板打电话来，说是没有SO8的，只有MSOP8封装的。得！这回还真要老命了，SO8的脚距是1.27毫米，MSOP8的脚距是0.65毫米，俺这老眼昏花滴就只好硬着头皮入瓮啦．．．

zzm530919

支持老同志，魂牵萦绕，这种感觉我也有过，不好受，买就买啦，玩他一下又何妨

风之澜

见到好宝贝俺也想把她牵回家

眉间尺

现在有各种小封装器件的小转大的板子卖.

e3po

ADI  的 AD8638 AD8639 也不错的

The AD8638/AD8639 are single and dual wide bandwidth,
auto-zero amplifiers featuring rail-to-rail output swing and low
noise. These amplifiers have very low offset, drift, and bias
current. Operation is fully specified from 5 V to 16 V single
supply (±2.5 V to ±8 V dual supply).
The AD8638/AD8639 provide benefits previously found only in expensive zero-drift or chopper-stabilized amplifiers. Using
the Analog Devices, Inc., topology, these auto-zero amplifiers
combine low cost with high accuracy and low noise. No exter-
nal capacitors are required. In addition, the AD8638/AD8639
greatly reduce the digital switching noise found in most chopper-
stabilized amplifiers.
With a typical offset voltage of only 3 μV, drift of 0.01 μV/°C,
and noise of 1.2 μV p-p (0.1 Hz to 10 Hz), the AD8638/AD8639
are suited for applications in which error sources cannot be
tolerated. Position and pressure sensors, medical equipment,
and strain gage amplifiers benefit greatly from nearly zero drift
over their operating temperature ranges. Many systems can take
advantage of the rail-to-rail output swing provided by
the AD8638/AD8639 to maximize signal-to-noise ratio (SNR).

e3po

原帖由 longshort 于 2010-9-19 17:08 发表
这一向一直在找一种价格和性能都能够承受的高精度运放，准备用于未来的电压参考源中。其实自己手中本来有几个国产的低漂移运放FC74，查参数也不错，失调温漂在0.5uV/摄氏度，但一则是体积稍大，同时使用几个的话均温 ...

OPA735 的原理在这里有讲到啊:


The OPA734 and OPA735 series of op amps use an
auto-zero topology with a time-continuous 1.6MHz op amp
in the signal path. This amplifier is zero-corrected every
100μs using a proprietary technique. Upon power-up, the
amplifier requires one full auto-zero cycle of approximately
100μs in addition to the start-up time for the bias circuitry
to achieve specified VOS accuracy. Prior to this time, the
amplifier may function properly but with unspecified offset
voltage.

Low-gain (< 20) operation demands that the auto-zero
circuitry correct for common-mode rejection errors of the
main amplifier. Because these errors can be larger than
0.1% of a full-scale input step change, one calibration
cycle (100μs) can be required to achieve full accuracy.
The term clock feedthrough describes the presence of the
clock frequency in the output spectrum. In auto-zeroed op
amps, clock feedthrough may result from the settling of the
internal sampling capacitor, or from the small amount of
charge injection that occurs during the sample-and-hold of
the op amp offset voltage. Feedthrough can be minimized
by keeping the source impedance relatively low (< 1kΩ)
and matching the source impedance on both input
terminals. If the source resistance is high (> 1kΩ)
feedthrough can generally be reduced with a capacitor of
1nF or greater in parallel with the source or feedback
resistors. See the circuit application examples.


根据以上描述， 大约可以猜测它也是斩波稳零， 不过不是老产品那种几十KHz 的频率， 而是 10M Hz 的频率。


这个问 millwood 可能可以得到比较满意的答案, 因为他/她就象电子学百科全书一样.

[ 本帖最后由 e3po 于 2010-9-19 20:23 编辑 ]

CGRCQZ

挡不住的诱惑

millwood

yes, they are simply the good old commutation (ac) amp, in a modern form.

10 years ago, they were the raged and expensive. but now, they are fairly common. many chip oems offer them. for example lmv2011 is widely available and has better drift performance than the  ti part.

those are primarily used for DC purposes, however, as their ac noise is significant.

而是 10M Hz 的频率。

I wouldn't bank on that, . the biggest give-away for those chips is their noise vs. frequency chart.

here is the one for the ti part. see the jump at 18khz and their after?

those parts are basically useless for audio.

longshort

谢谢各位解答！谢谢milwood！

确实，从特性图上看噪声水平确实比较高，不过用于直流应用还是适合的。从厂方提供的电路来看，电容器的全负反馈是主要形式。没有这个负反馈，10Hz带宽内的噪声峰峰值可以达到3uV左右，超过失调温漂的300倍。这个噪声的来源，基本上就是稳零系统造成的取样噪声。

俺想，用于将电压基准传递到输出，这个产品还是能够胜任的。对于一个温度系数为正负5ppm/摄氏度的6.2V参考源来说，运放的失调温漂引起的参考源总温度系数变动，基本上可以忽略不计，前提是它们都位于同一级温度梯度线上。

longshort

原帖由 e3po 于 2010-9-19 20:00 发表
ADI  的 AD8638 AD8639 也不错的

The AD8638/AD8639 are single and dual wide bandwidth,
auto-zero amplifiers featuring rail-to-rail output swing and low
noise. These amplifiers have very low offset, ...

参数确实诱人，比OPA735稍好，不过价格比较高，查了下淘宝，比较一致的价格在25~35元之间。

longshort

原帖由 眉间尺 于 2010-9-19 19:50 发表
现在有各种小封装器件的小转大的板子卖.

俺想自己做个双列直插的板子，等东西到了把它们焊上，就成了DIP产品了。

e3po

原帖由 millwood 于 2010-9-19 20:56 发表
yes, they are simply the good old commutation (ac) amp, in a modern form.

10 years ago, they were the raged and expensive. but now, they are fairly common. many chip oems offer them. for example lm ...

That 10MHz thing is my conjecture,  I'd been hoping it were a good guess.

BTW,

Is the AD8638/8639 any good for building a ECG/EKG preamp ?

[ 本帖最后由 e3po 于 2010-9-19 22:05 编辑 ]

longshort

原帖由 e3po 于 2010-9-19 20:18 发表



OPA735 的原理在这里有讲到啊:


The OPA734 and OPA735 series of op amps use an
auto-zero topology with a time-continuous 1.6MHz op amp
in the signal path. This amplifier is zero-corrected every
100μs using a proprietary technique. Upon power-up, the
amplifier requires one full auto-zero cycle of approximately
100μs in addition to the start-up time for the bias circuitry
to achieve specified VOS accuracy. Prior to this time, the
amplifier may function properly but with unspecified offset
voltage.

Low-gain (< 20) operation demands that the auto-zero
circuitry correct for common-mode rejection errors of the
main amplifier. Because these errors can be larger than
0.1% of a full-scale input step change, one calibration
cycle (100μs) can be required to achieve full accuracy.
The term clock feedthrough describes the presence of the
clock frequency in the output spectrum. In auto-zeroed op
amps, clock feedthrough may result from the settling of the
internal sampling capacitor, or from the small amount of
charge injection that occurs during the sample-and-hold of
the op amp offset voltage. Feedthrough can be minimized
by keeping the source impedance relatively low (< 1kΩ)
and matching the source impedance on both input
terminals. If the source resistance is high (> 1kΩ)
feedthrough can generally be reduced with a capacitor of
1nF or greater in parallel with the source or feedback
resistors. See the circuit application examples.


根据以上描述， 大约可以猜测它也是斩波稳零， 不过不是老产品那种几十KHz 的频率， 而是 10M Hz 的频率。

俺看了半天，没看到10MHz，倒是看到了这么一句：“one calibration cycle (100μs) can be required to achieve full accuracy.”。算起来好象是10KHz？

e3po

原帖由 longshort 于 2010-9-19 22:02 发表

俺看了半天，没看到10MHz，倒是看到了这么一句：“one calibration cycle (100μs) can be required to achieve full accuracy.”。算起来好象是10KHz？

            咳咳，岁月不饶人啊~~~~

          吾也老了，  一眼扫过去还以为是 100nS.

longshort

原帖由 zzm530919 于 2010-9-19 17:19 发表
支持老同志，魂牵萦绕，这种感觉我也有过，不好受，买就买啦，玩他一下又何妨

嘿嘿，老骨头要尝鲜，滋味一定不错！